223
|
1 /* vi:set ts=8 sts=4 sw=4:
|
|
2 *
|
|
3 * VIM - Vi IMproved by Bram Moolenaar
|
|
4 *
|
|
5 * Do ":help uganda" in Vim to read copying and usage conditions.
|
|
6 * Do ":help credits" in Vim to see a list of people who contributed.
|
|
7 * See README.txt for an overview of the Vim source code.
|
|
8 */
|
|
9
|
|
10 /*
|
|
11 * spell.c: code for spell checking
|
226
|
12 *
|
300
|
13 * The spell checking mechanism uses a tree (aka trie). Each node in the tree
|
|
14 * has a list of bytes that can appear (siblings). For each byte there is a
|
|
15 * pointer to the node with the byte that follows in the word (child).
|
|
16 * A NUL byte is used where the word may end.
|
|
17 *
|
|
18 * There are two trees: one with case-folded words and one with words in
|
|
19 * original case. The second one is only used for keep-case words and is
|
|
20 * usually small.
|
|
21 *
|
|
22 * Thanks to Olaf Seibert for providing an example implementation of this tree
|
|
23 * and the compression mechanism.
|
243
|
24 *
|
|
25 * Matching involves checking the caps type: Onecap ALLCAP KeepCap.
|
|
26 *
|
236
|
27 * Why doesn't Vim use aspell/ispell/myspell/etc.?
|
|
28 * See ":help develop-spell".
|
|
29 */
|
|
30
|
300
|
31 /*
|
323
|
32 * Use this to let the score depend in how much a suggestion sounds like the
|
|
33 * bad word. It's quite slow and doesn't make the sorting much better....
|
|
34 * #define SOUNDFOLD_SCORE
|
|
35 */
|
|
36
|
|
37 /*
|
300
|
38 * Vim spell file format: <HEADER> <SUGGEST> <LWORDTREE> <KWORDTREE>
|
|
39 *
|
|
40 * <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
41 * <charflagslen> <charflags> <fcharslen> <fchars>
|
|
42 *
|
323
|
43 * <fileID> 10 bytes "VIMspell06"
|
300
|
44 * <regioncnt> 1 byte number of regions following (8 supported)
|
307
|
45 * <regionname> 2 bytes Region name: ca, au, etc. Lower case.
|
300
|
46 * First <regionname> is region 1.
|
|
47 *
|
|
48 * <charflagslen> 1 byte Number of bytes in <charflags> (should be 128).
|
|
49 * <charflags> N bytes List of flags (first one is for character 128):
|
|
50 * 0x01 word character
|
307
|
51 * 0x02 upper-case character
|
300
|
52 * <fcharslen> 2 bytes Number of bytes in <fchars>.
|
|
53 * <fchars> N bytes Folded characters, first one is for character 128.
|
|
54 *
|
|
55 *
|
323
|
56 * <SUGGEST> : <repcount> <rep> ...
|
|
57 * <salflags> <salcount> <sal> ...
|
|
58 * <maplen> <mapstr>
|
|
59 *
|
|
60 * <repcount> 2 bytes number of <rep> items, MSB first.
|
|
61 *
|
|
62 * <rep> : <repfromlen> <repfrom> <reptolen> <repto>
|
|
63 *
|
|
64 * <repfromlen> 1 byte length of <repfrom>
|
|
65 *
|
|
66 * <repfrom> N bytes "from" part of replacement
|
|
67 *
|
|
68 * <reptolen> 1 byte length of <repto>
|
|
69 *
|
|
70 * <repto> N bytes "to" part of replacement
|
300
|
71 *
|
323
|
72 * <salflags> 1 byte flags for soundsalike conversion:
|
|
73 * SAL_F0LLOWUP
|
|
74 * SAL_COLLAPSE
|
|
75 * SAL_REM_ACCENTS
|
|
76 *
|
|
77 * <sal> : <salfromlen> <salfrom> <saltolen> <salto>
|
|
78 *
|
|
79 * <salfromlen> 1 byte length of <salfrom>
|
|
80 *
|
|
81 * <salfrom> N bytes "from" part of soundsalike
|
|
82 *
|
|
83 * <saltolen> 1 byte length of <salto>
|
|
84 *
|
|
85 * <salto> N bytes "to" part of soundsalike
|
|
86 *
|
|
87 * <maplen> 2 bytes length of <mapstr>, MSB first
|
|
88 *
|
|
89 * <mapstr> N bytes String with sequences of similar characters,
|
|
90 * separated by slashes.
|
300
|
91 *
|
|
92 *
|
|
93 * <LWORDTREE>: <wordtree>
|
|
94 *
|
|
95 * <wordtree>: <nodecount> <nodedata> ...
|
|
96 *
|
|
97 * <nodecount> 4 bytes Number of nodes following. MSB first.
|
|
98 *
|
|
99 * <nodedata>: <siblingcount> <sibling> ...
|
|
100 *
|
|
101 * <siblingcount> 1 byte Number of siblings in this node. The siblings
|
|
102 * follow in sorted order.
|
|
103 *
|
|
104 * <sibling>: <byte> [<nodeidx> <xbyte> | <flags> [<region>]]
|
|
105 *
|
|
106 * <byte> 1 byte Byte value of the sibling. Special cases:
|
|
107 * BY_NOFLAGS: End of word without flags and for all
|
|
108 * regions.
|
|
109 * BY_FLAGS: End of word, <flags> follow.
|
|
110 * BY_INDEX: Child of sibling is shared, <nodeidx>
|
|
111 * and <xbyte> follow.
|
|
112 *
|
|
113 * <nodeidx> 3 bytes Index of child for this sibling, MSB first.
|
|
114 *
|
|
115 * <xbyte> 1 byte byte value of the sibling.
|
|
116 *
|
|
117 * <flags> 1 byte bitmask of:
|
|
118 * WF_ALLCAP word must have only capitals
|
|
119 * WF_ONECAP first char of word must be capital
|
|
120 * WF_RARE rare word
|
|
121 * WF_REGION <region> follows
|
|
122 *
|
|
123 * <region> 1 byte Bitmask for regions in which word is valid. When
|
|
124 * omitted it's valid in all regions.
|
|
125 * Lowest bit is for region 1.
|
|
126 *
|
|
127 * <KWORDTREE>: <wordtree>
|
|
128 *
|
|
129 * All text characters are in 'encoding', but stored as single bytes.
|
|
130 */
|
|
131
|
223
|
132 #if defined(MSDOS) || defined(WIN16) || defined(WIN32) || defined(_WIN64)
|
|
133 # include <io.h> /* for lseek(), must be before vim.h */
|
|
134 #endif
|
|
135
|
|
136 #include "vim.h"
|
|
137
|
|
138 #if defined(FEAT_SYN_HL) || defined(PROTO)
|
|
139
|
|
140 #ifdef HAVE_FCNTL_H
|
|
141 # include <fcntl.h>
|
|
142 #endif
|
|
143
|
323
|
144 #define MAXWLEN 250 /* Assume max. word len is this many bytes.
|
|
145 Some places assume a word length fits in a
|
|
146 byte, thus it can't be above 255. */
|
226
|
147
|
300
|
148 /* Flags used for a word. */
|
|
149 #define WF_REGION 0x01 /* region byte follows */
|
|
150 #define WF_ONECAP 0x02 /* word with one capital (or all capitals) */
|
|
151 #define WF_ALLCAP 0x04 /* word must be all capitals */
|
|
152 #define WF_RARE 0x08 /* rare word */
|
307
|
153 #define WF_BANNED 0x10 /* bad word */
|
323
|
154 #define WF_KEEPCAP 0x80 /* keep-case word */
|
|
155
|
|
156 #define WF_CAPMASK (WF_ONECAP | WF_ALLCAP | WF_KEEPCAP)
|
300
|
157
|
|
158 #define BY_NOFLAGS 0 /* end of word without flags or region */
|
|
159 #define BY_FLAGS 1 /* end of word, flag byte follows */
|
|
160 #define BY_INDEX 2 /* child is shared, index follows */
|
|
161 #define BY_SPECIAL BY_INDEX /* hightest special byte value */
|
236
|
162
|
323
|
163 /* Info from "REP" and "SAL" entries in ".aff" file used in si_rep, sl_rep,
|
|
164 * si_sal and sl_sal.
|
|
165 * One replacement: from "ft_from" to "ft_to". */
|
|
166 typedef struct fromto_S
|
236
|
167 {
|
323
|
168 char_u *ft_from;
|
|
169 char_u *ft_to;
|
|
170 } fromto_T;
|
236
|
171
|
|
172 /*
|
243
|
173 * Structure used to store words and other info for one language, loaded from
|
|
174 * a .spl file.
|
300
|
175 * The main access is through the tree in "sl_fbyts/sl_fidxs", storing the
|
|
176 * case-folded words. "sl_kbyts/sl_kidxs" is for keep-case words.
|
|
177 *
|
|
178 * The "byts" array stores the possible bytes in each tree node, preceded by
|
|
179 * the number of possible bytes, sorted on byte value:
|
|
180 * <len> <byte1> <byte2> ...
|
|
181 * The "idxs" array stores the index of the child node corresponding to the
|
|
182 * byte in "byts".
|
|
183 * Exception: when the byte is zero, the word may end here and "idxs" holds
|
|
184 * the flags and region for the word. There may be several zeros in sequence
|
|
185 * for alternative flag/region combinations.
|
236
|
186 */
|
|
187 typedef struct slang_S slang_T;
|
|
188 struct slang_S
|
|
189 {
|
|
190 slang_T *sl_next; /* next language */
|
|
191 char_u *sl_name; /* language name "en", "en.rare", "nl", etc. */
|
310
|
192 char_u *sl_fname; /* name of .spl file */
|
323
|
193 int sl_add; /* TRUE if it's a .add file. */
|
300
|
194 char_u *sl_fbyts; /* case-folded word bytes */
|
|
195 int *sl_fidxs; /* case-folded word indexes */
|
|
196 char_u *sl_kbyts; /* keep-case word bytes */
|
|
197 int *sl_kidxs; /* keep-case word indexes */
|
236
|
198 char_u sl_regions[17]; /* table with up to 8 region names plus NUL */
|
323
|
199
|
|
200 garray_T sl_rep; /* list of fromto_T entries from REP lines */
|
|
201 short sl_rep_first[256]; /* indexes where byte first appears, -1 if
|
|
202 there is none */
|
|
203 garray_T sl_sal; /* list of fromto_T entries from SAL lines */
|
|
204 short sl_sal_first[256]; /* indexes where byte first appears, -1 if
|
|
205 there is none */
|
|
206 int sl_followup; /* SAL followup */
|
|
207 int sl_collapse; /* SAL collapse_result */
|
|
208 int sl_rem_accents; /* SAL remove_accents */
|
|
209 char_u *sl_map; /* string with similar chars from MAP lines */
|
236
|
210 };
|
|
211
|
243
|
212 /* First language that is loaded, start of the linked list of loaded
|
|
213 * languages. */
|
236
|
214 static slang_T *first_lang = NULL;
|
|
215
|
323
|
216 /* Flags used in .spl file for soundsalike flags. */
|
|
217 #define SAL_F0LLOWUP 1
|
|
218 #define SAL_COLLAPSE 2
|
|
219 #define SAL_REM_ACCENTS 4
|
|
220
|
236
|
221 /*
|
|
222 * Structure used in "b_langp", filled from 'spelllang'.
|
|
223 */
|
|
224 typedef struct langp_S
|
|
225 {
|
|
226 slang_T *lp_slang; /* info for this language (NULL for last one) */
|
|
227 int lp_region; /* bitmask for region or REGION_ALL */
|
|
228 } langp_T;
|
|
229
|
|
230 #define LANGP_ENTRY(ga, i) (((langp_T *)(ga).ga_data) + (i))
|
|
231
|
307
|
232 #define REGION_ALL 0xff /* word valid in all regions */
|
|
233
|
|
234 /* Result values. Lower number is accepted over higher one. */
|
|
235 #define SP_BANNED -1
|
236
|
236 #define SP_OK 0
|
307
|
237 #define SP_RARE 1
|
|
238 #define SP_LOCAL 2
|
|
239 #define SP_BAD 3
|
236
|
240
|
323
|
241 #define VIMSPELLMAGIC "VIMspell06" /* string at start of Vim spell file */
|
236
|
242 #define VIMSPELLMAGICL 10
|
|
243
|
|
244 /*
|
323
|
245 * Information used when looking for suggestions.
|
|
246 */
|
|
247 typedef struct suginfo_S
|
|
248 {
|
|
249 garray_T su_ga; /* suggestions, contains "suggest_T" */
|
|
250 int su_maxscore; /* maximum score for adding to su_ga */
|
|
251 int su_icase; /* accept words with wrong case */
|
|
252 int su_icase_add; /* add matches while ignoring case */
|
|
253 char_u *su_badptr; /* start of bad word in line */
|
|
254 int su_badlen; /* length of detected bad word in line */
|
|
255 char_u su_badword[MAXWLEN]; /* bad word truncated at su_badlen */
|
|
256 char_u su_fbadword[MAXWLEN]; /* su_badword case-folded */
|
|
257 hashtab_T su_banned; /* table with banned words */
|
|
258 #ifdef SOUNDFOLD_SCORE
|
|
259 slang_T *su_slang; /* currently used slang_T */
|
|
260 char_u su_salword[MAXWLEN]; /* soundfolded badword */
|
|
261 #endif
|
|
262 } suginfo_T;
|
|
263
|
|
264 /* One word suggestion. Used in "si_ga". */
|
|
265 typedef struct suggest_S
|
|
266 {
|
|
267 char_u *st_word; /* suggested word, allocated string */
|
|
268 int st_orglen; /* length of replaced text */
|
|
269 int st_score; /* lower is better */
|
|
270 } suggest_T;
|
|
271
|
|
272 #define SUG(sup, i) (((suggest_T *)(sup)->su_ga.ga_data)[i])
|
|
273
|
|
274 /* Number of suggestions displayed. */
|
|
275 #define SUG_PROMPT_COUNT ((int)Rows - 2)
|
|
276
|
|
277 /* Threshold for sorting and cleaning up suggestions. */
|
|
278 #define SUG_CLEANUP_COUNT (SUG_PROMPT_COUNT + 50)
|
|
279
|
|
280 /* score for various changes */
|
|
281 #define SCORE_SPLIT 99 /* split bad word */
|
|
282 #define SCORE_ICASE 52 /* slightly different case */
|
|
283 #define SCORE_ALLCAP 120 /* need all-cap case */
|
|
284 #define SCORE_REGION 70 /* word is for different region */
|
|
285 #define SCORE_RARE 180 /* rare word */
|
|
286
|
|
287 /* score for edit distance */
|
|
288 #define SCORE_SWAP 90 /* swap two characters */
|
|
289 #define SCORE_SWAP3 110 /* swap two characters in three */
|
|
290 #define SCORE_REP 87 /* REP replacement */
|
|
291 #define SCORE_SUBST 93 /* substitute a character */
|
|
292 #define SCORE_SIMILAR 33 /* substitute a similar character */
|
|
293 #define SCORE_DEL 96 /* delete a character */
|
|
294 #define SCORE_INS 94 /* insert a character */
|
|
295
|
|
296 #define SCORE_MAXINIT 350 /* Initial maximum score: higher == slower.
|
|
297 * 350 allows for about three changes. */
|
|
298 #define SCORE_MAXMAX 999999 /* accept any score */
|
|
299
|
|
300 /*
|
236
|
301 * Structure to store info for word matching.
|
|
302 */
|
|
303 typedef struct matchinf_S
|
|
304 {
|
|
305 langp_T *mi_lp; /* info for language and region */
|
243
|
306
|
|
307 /* pointers to original text to be checked */
|
236
|
308 char_u *mi_word; /* start of word being checked */
|
300
|
309 char_u *mi_end; /* end of matching word */
|
243
|
310 char_u *mi_fend; /* next char to be added to mi_fword */
|
300
|
311 char_u *mi_cend; /* char after what was used for
|
|
312 mi_capflags */
|
243
|
313
|
|
314 /* case-folded text */
|
|
315 char_u mi_fword[MAXWLEN + 1]; /* mi_word case-folded */
|
300
|
316 int mi_fwordlen; /* nr of valid bytes in mi_fword */
|
243
|
317
|
|
318 /* others */
|
236
|
319 int mi_result; /* result so far: SP_BAD, SP_OK, etc. */
|
300
|
320 int mi_capflags; /* WF_ONECAP WF_ALLCAP WF_KEEPCAP */
|
236
|
321 } matchinf_T;
|
|
322
|
307
|
323 /*
|
|
324 * The tables used for recognizing word characters according to spelling.
|
|
325 * These are only used for the first 256 characters of 'encoding'.
|
|
326 */
|
|
327 typedef struct spelltab_S
|
|
328 {
|
|
329 char_u st_isw[256]; /* flags: is word char */
|
|
330 char_u st_isu[256]; /* flags: is uppercase char */
|
|
331 char_u st_fold[256]; /* chars: folded case */
|
|
332 } spelltab_T;
|
|
333
|
|
334 static spelltab_T spelltab;
|
|
335 static int did_set_spelltab;
|
|
336
|
|
337 #define SPELL_ISWORD 1
|
|
338 #define SPELL_ISUPPER 2
|
|
339
|
|
340 static void clear_spell_chartab __ARGS((spelltab_T *sp));
|
|
341 static int set_spell_finish __ARGS((spelltab_T *new_st));
|
|
342
|
|
343 /*
|
|
344 * Return TRUE if "p" points to a word character or "c" is a word character
|
|
345 * for spelling.
|
|
346 * Checking for a word character is done very often, avoid the function call
|
|
347 * overhead.
|
|
348 */
|
|
349 #ifdef FEAT_MBYTE
|
|
350 # define SPELL_ISWORDP(p) ((has_mbyte && MB_BYTE2LEN(*(p)) > 1) \
|
|
351 ? (mb_get_class(p) >= 2) : spelltab.st_isw[*(p)])
|
|
352 #else
|
|
353 # define SPELL_ISWORDP(p) (spelltab.st_isw[*(p)])
|
|
354 #endif
|
|
355
|
323
|
356 /*
|
|
357 * Struct to keep the state at each level in spell_try_change().
|
|
358 */
|
|
359 typedef struct trystate_S
|
|
360 {
|
|
361 int ts_state; /* state at this level, STATE_ */
|
|
362 int ts_score; /* score */
|
|
363 int ts_curi; /* index in list of child nodes */
|
|
364 int ts_fidx; /* index in fword[], case-folded bad word */
|
|
365 int ts_fidxtry; /* ts_fidx at which bytes may be changed */
|
|
366 int ts_twordlen; /* valid length of tword[] */
|
|
367 int ts_arridx; /* index in tree array, start of node */
|
|
368 char_u ts_save_prewordlen; /* saved "prewordlen" */
|
|
369 int ts_save_splitoff; /* su_splitoff saved here */
|
|
370 int ts_save_badflags; /* badflags saved here */
|
|
371 } trystate_T;
|
|
372
|
236
|
373 static slang_T *slang_alloc __ARGS((char_u *lang));
|
|
374 static void slang_free __ARGS((slang_T *lp));
|
310
|
375 static void slang_clear __ARGS((slang_T *lp));
|
300
|
376 static void find_word __ARGS((matchinf_T *mip, int keepcap));
|
323
|
377 static int spell_valid_case __ARGS((int origflags, int treeflags));
|
307
|
378 static void spell_load_lang __ARGS((char_u *lang));
|
310
|
379 static char_u *spell_enc __ARGS((void));
|
|
380 static void spell_load_cb __ARGS((char_u *fname, void *cookie));
|
323
|
381 static slang_T *spell_load_file __ARGS((char_u *fname, char_u *lang, slang_T *old_lp, int silent));
|
300
|
382 static int read_tree __ARGS((FILE *fd, char_u *byts, int *idxs, int maxidx, int startidx));
|
236
|
383 static int find_region __ARGS((char_u *rp, char_u *region));
|
|
384 static int captype __ARGS((char_u *word, char_u *end));
|
323
|
385 static void spell_reload_one __ARGS((char_u *fname, int added_word));
|
307
|
386 static int set_spell_charflags __ARGS((char_u *flags, int cnt, char_u *upp));
|
|
387 static int set_spell_chartab __ARGS((char_u *fol, char_u *low, char_u *upp));
|
|
388 static void write_spell_chartab __ARGS((FILE *fd));
|
|
389 static int spell_isupper __ARGS((int c));
|
|
390 static int spell_casefold __ARGS((char_u *p, int len, char_u *buf, int buflen));
|
323
|
391 static void onecap_copy __ARGS((char_u *word, int len, char_u *wcopy, int upper));
|
|
392 static void spell_try_change __ARGS((suginfo_T *su));
|
|
393 static int try_deeper __ARGS((suginfo_T *su, trystate_T *stack, int depth, int score_add));
|
|
394 static void find_keepcap_word __ARGS((slang_T *slang, char_u *fword, char_u *kword));
|
|
395 static void spell_try_soundalike __ARGS((suginfo_T *su));
|
|
396 static void make_case_word __ARGS((char_u *fword, char_u *cword, int flags));
|
|
397 static int similar_chars __ARGS((slang_T *slang, int c1, int c2));
|
|
398 static void add_suggestion __ARGS((suginfo_T *su, char_u *goodword, int use_score));
|
|
399 static void add_banned __ARGS((suginfo_T *su, char_u *word));
|
|
400 static int was_banned __ARGS((suginfo_T *su, char_u *word));
|
|
401 static void free_banned __ARGS((suginfo_T *su));
|
|
402 static void cleanup_suggestions __ARGS((suginfo_T *su));
|
|
403 static void spell_soundfold __ARGS((slang_T *slang, char_u *inword, char_u *res));
|
|
404 static int spell_edit_score __ARGS((char_u *badword, char_u *goodword));
|
|
405
|
307
|
406
|
|
407 static char *e_format = N_("E759: Format error in spell file");
|
236
|
408
|
|
409 /*
|
|
410 * Main spell-checking function.
|
300
|
411 * "ptr" points to a character that could be the start of a word.
|
236
|
412 * "*attrp" is set to the attributes for a badly spelled word. For a non-word
|
|
413 * or when it's OK it remains unchanged.
|
|
414 * This must only be called when 'spelllang' is not empty.
|
323
|
415 *
|
|
416 * "sug" is normally NULL. When looking for suggestions it points to
|
|
417 * suginfo_T. It's passed as a void pointer to keep the struct local.
|
|
418 *
|
236
|
419 * Returns the length of the word in bytes, also when it's OK, so that the
|
|
420 * caller can skip over the word.
|
|
421 */
|
|
422 int
|
300
|
423 spell_check(wp, ptr, attrp)
|
236
|
424 win_T *wp; /* current window */
|
|
425 char_u *ptr;
|
|
426 int *attrp;
|
|
427 {
|
|
428 matchinf_T mi; /* Most things are put in "mi" so that it can
|
|
429 be passed to functions quickly. */
|
|
430
|
307
|
431 /* A word never starts at a space or a control character. Return quickly
|
|
432 * then, skipping over the character. */
|
|
433 if (*ptr <= ' ')
|
|
434 return 1;
|
236
|
435
|
300
|
436 /* A word starting with a number is always OK. Also skip hexadecimal
|
|
437 * numbers 0xFF99 and 0X99FF. */
|
|
438 if (*ptr >= '0' && *ptr <= '9')
|
|
439 {
|
316
|
440 if (*ptr == '0' && (ptr[1] == 'x' || ptr[1] == 'X'))
|
|
441 mi.mi_end = skiphex(ptr + 2);
|
300
|
442 else
|
|
443 mi.mi_end = skipdigits(ptr);
|
|
444 }
|
|
445 else
|
236
|
446 {
|
307
|
447 /* Find the end of the word. */
|
|
448 mi.mi_word = ptr;
|
300
|
449 mi.mi_fend = ptr;
|
323
|
450
|
307
|
451 if (SPELL_ISWORDP(mi.mi_fend))
|
300
|
452 {
|
|
453 /* Make case-folded copy of the characters until the next non-word
|
|
454 * character. */
|
|
455 do
|
|
456 {
|
|
457 mb_ptr_adv(mi.mi_fend);
|
307
|
458 } while (*mi.mi_fend != NUL && SPELL_ISWORDP(mi.mi_fend));
|
300
|
459 }
|
307
|
460
|
|
461 /* We always use the characters up to the next non-word character,
|
|
462 * also for bad words. */
|
|
463 mi.mi_end = mi.mi_fend;
|
323
|
464
|
|
465 /* Check caps type later. */
|
|
466 mi.mi_capflags = 0;
|
|
467 mi.mi_cend = NULL;
|
300
|
468
|
307
|
469 /* Include one non-word character so that we can check for the
|
|
470 * word end. */
|
|
471 if (*mi.mi_fend != NUL)
|
|
472 mb_ptr_adv(mi.mi_fend);
|
|
473
|
|
474 (void)spell_casefold(ptr, (int)(mi.mi_fend - ptr), mi.mi_fword,
|
|
475 MAXWLEN + 1);
|
|
476 mi.mi_fwordlen = STRLEN(mi.mi_fword);
|
|
477
|
300
|
478 /* The word is bad unless we recognize it. */
|
|
479 mi.mi_result = SP_BAD;
|
236
|
480
|
300
|
481 /*
|
|
482 * Loop over the languages specified in 'spelllang'.
|
|
483 * We check them all, because a matching word may be longer than an
|
|
484 * already found matching word.
|
|
485 */
|
|
486 for (mi.mi_lp = LANGP_ENTRY(wp->w_buffer->b_langp, 0);
|
|
487 mi.mi_lp->lp_slang != NULL; ++mi.mi_lp)
|
243
|
488 {
|
300
|
489 /* Check for a matching word in case-folded words. */
|
|
490 find_word(&mi, FALSE);
|
|
491
|
|
492 find_word(&mi, TRUE);
|
|
493 }
|
243
|
494
|
300
|
495 if (mi.mi_result != SP_OK)
|
|
496 {
|
|
497 /* When we are at a non-word character there is no error, just
|
|
498 * skip over the character (try looking for a word after it). */
|
307
|
499 if (!SPELL_ISWORDP(ptr))
|
243
|
500 {
|
300
|
501 #ifdef FEAT_MBYTE
|
|
502 if (has_mbyte)
|
|
503 return mb_ptr2len_check(ptr);
|
|
504 #endif
|
|
505 return 1;
|
243
|
506 }
|
|
507
|
307
|
508 if (mi.mi_result == SP_BAD || mi.mi_result == SP_BANNED)
|
300
|
509 *attrp = highlight_attr[HLF_SPB];
|
|
510 else if (mi.mi_result == SP_RARE)
|
|
511 *attrp = highlight_attr[HLF_SPR];
|
|
512 else
|
|
513 *attrp = highlight_attr[HLF_SPL];
|
243
|
514 }
|
|
515 }
|
|
516
|
300
|
517 return (int)(mi.mi_end - ptr);
|
236
|
518 }
|
|
519
|
|
520 /*
|
300
|
521 * Check if the word at "mip->mi_word" is in the tree.
|
|
522 * When "keepcap" is TRUE check in keep-case word tree.
|
|
523 *
|
|
524 * For a match mip->mi_result is updated.
|
243
|
525 */
|
|
526 static void
|
300
|
527 find_word(mip, keepcap)
|
243
|
528 matchinf_T *mip;
|
300
|
529 int keepcap;
|
243
|
530 {
|
300
|
531 int arridx = 0;
|
|
532 int endlen[MAXWLEN]; /* length at possible word endings */
|
|
533 int endidx[MAXWLEN]; /* possible word endings */
|
|
534 int endidxcnt = 0;
|
|
535 int len;
|
|
536 int wlen = 0;
|
|
537 int flen;
|
|
538 int c;
|
|
539 char_u *ptr;
|
|
540 unsigned lo, hi, m;
|
243
|
541 #ifdef FEAT_MBYTE
|
300
|
542 char_u *s;
|
307
|
543 #endif
|
300
|
544 char_u *p;
|
307
|
545 int res = SP_BAD;
|
|
546 int valid;
|
300
|
547 slang_T *slang = mip->mi_lp->lp_slang;
|
|
548 unsigned flags;
|
|
549 char_u *byts;
|
|
550 int *idxs;
|
243
|
551
|
300
|
552 if (keepcap)
|
236
|
553 {
|
300
|
554 /* Check for word with matching case in keep-case tree. */
|
|
555 ptr = mip->mi_word;
|
|
556 flen = 9999; /* no case folding, always enough bytes */
|
|
557 byts = slang->sl_kbyts;
|
|
558 idxs = slang->sl_kidxs;
|
236
|
559 }
|
|
560 else
|
|
561 {
|
300
|
562 /* Check for case-folded in case-folded tree. */
|
|
563 ptr = mip->mi_fword;
|
|
564 flen = mip->mi_fwordlen; /* available case-folded bytes */
|
|
565 byts = slang->sl_fbyts;
|
|
566 idxs = slang->sl_fidxs;
|
243
|
567 }
|
|
568
|
300
|
569 if (byts == NULL)
|
|
570 return; /* array is empty */
|
236
|
571
|
|
572 /*
|
307
|
573 * Repeat advancing in the tree until:
|
|
574 * - there is a byte that doesn't match,
|
|
575 * - we reach the end of the tree,
|
|
576 * - or we reach the end of the line.
|
236
|
577 */
|
300
|
578 for (;;)
|
236
|
579 {
|
300
|
580 if (flen == 0 && *mip->mi_fend != NUL)
|
236
|
581 {
|
300
|
582 /* Need to fold at least one more character. Do until next
|
|
583 * non-word character for efficiency. */
|
307
|
584 p = mip->mi_fend;
|
300
|
585 do
|
236
|
586 {
|
307
|
587 mb_ptr_adv(mip->mi_fend);
|
|
588 } while (*mip->mi_fend != NUL && SPELL_ISWORDP(mip->mi_fend));
|
|
589
|
|
590 /* Include the non-word character so that we can check for the
|
|
591 * word end. */
|
|
592 if (*mip->mi_fend != NUL)
|
|
593 mb_ptr_adv(mip->mi_fend);
|
|
594
|
|
595 (void)spell_casefold(p, (int)(mip->mi_fend - p),
|
300
|
596 mip->mi_fword + mip->mi_fwordlen,
|
|
597 MAXWLEN - mip->mi_fwordlen);
|
|
598 flen = STRLEN(mip->mi_fword + mip->mi_fwordlen);
|
|
599 mip->mi_fwordlen += flen;
|
|
600 }
|
|
601
|
|
602 len = byts[arridx++];
|
|
603
|
|
604 /* If the first possible byte is a zero the word could end here.
|
|
605 * Remember this index, we first check for the longest word. */
|
|
606 if (byts[arridx] == 0)
|
|
607 {
|
307
|
608 if (endidxcnt == MAXWLEN)
|
|
609 {
|
|
610 /* Must be a corrupted spell file. */
|
|
611 EMSG(_(e_format));
|
|
612 return;
|
|
613 }
|
300
|
614 endlen[endidxcnt] = wlen;
|
|
615 endidx[endidxcnt++] = arridx++;
|
|
616 --len;
|
|
617
|
|
618 /* Skip over the zeros, there can be several flag/region
|
|
619 * combinations. */
|
|
620 while (len > 0 && byts[arridx] == 0)
|
|
621 {
|
|
622 ++arridx;
|
|
623 --len;
|
|
624 }
|
|
625 if (len == 0)
|
|
626 break; /* no children, word must end here */
|
|
627 }
|
|
628
|
|
629 /* Stop looking at end of the line. */
|
|
630 if (ptr[wlen] == NUL)
|
|
631 break;
|
|
632
|
|
633 /* Perform a binary search in the list of accepted bytes. */
|
|
634 c = ptr[wlen];
|
|
635 lo = arridx;
|
|
636 hi = arridx + len - 1;
|
|
637 while (lo < hi)
|
|
638 {
|
|
639 m = (lo + hi) / 2;
|
|
640 if (byts[m] > c)
|
|
641 hi = m - 1;
|
|
642 else if (byts[m] < c)
|
|
643 lo = m + 1;
|
|
644 else
|
|
645 {
|
|
646 lo = hi = m;
|
|
647 break;
|
236
|
648 }
|
|
649 }
|
300
|
650
|
|
651 /* Stop if there is no matching byte. */
|
|
652 if (hi < lo || byts[lo] != c)
|
|
653 break;
|
|
654
|
|
655 /* Continue at the child (if there is one). */
|
|
656 arridx = idxs[lo];
|
|
657 ++wlen;
|
|
658 --flen;
|
236
|
659 }
|
|
660
|
300
|
661 /*
|
|
662 * Verify that one of the possible endings is valid. Try the longest
|
|
663 * first.
|
|
664 */
|
|
665 while (endidxcnt > 0)
|
|
666 {
|
|
667 --endidxcnt;
|
|
668 arridx = endidx[endidxcnt];
|
|
669 wlen = endlen[endidxcnt];
|
236
|
670
|
300
|
671 #ifdef FEAT_MBYTE
|
|
672 if ((*mb_head_off)(ptr, ptr + wlen) > 0)
|
|
673 continue; /* not at first byte of character */
|
|
674 #endif
|
307
|
675 if (SPELL_ISWORDP(ptr + wlen))
|
300
|
676 continue; /* next char is a word character */
|
|
677
|
|
678 #ifdef FEAT_MBYTE
|
|
679 if (!keepcap && has_mbyte)
|
|
680 {
|
|
681 /* Compute byte length in original word, length may change
|
|
682 * when folding case. */
|
|
683 p = mip->mi_word;
|
|
684 for (s = ptr; s < ptr + wlen; mb_ptr_adv(s))
|
|
685 mb_ptr_adv(p);
|
|
686 wlen = p - mip->mi_word;
|
|
687 }
|
|
688 #endif
|
236
|
689
|
300
|
690 /* Check flags and region. Repeat this if there are more
|
|
691 * flags/region alternatives until there is a match. */
|
|
692 for (len = byts[arridx - 1]; len > 0 && byts[arridx] == 0; --len)
|
|
693 {
|
|
694 flags = idxs[arridx];
|
|
695 if (keepcap)
|
|
696 {
|
|
697 /* For "keepcap" tree the case is always right. */
|
|
698 valid = TRUE;
|
|
699 }
|
|
700 else
|
|
701 {
|
|
702 /* Check that the word is in the required case. */
|
|
703 if (mip->mi_cend != mip->mi_word + wlen)
|
|
704 {
|
323
|
705 /* mi_capflags was set for a different word length, need
|
|
706 * to do it again. */
|
300
|
707 mip->mi_cend = mip->mi_word + wlen;
|
323
|
708 mip->mi_capflags = captype(mip->mi_word, mip->mi_cend);
|
300
|
709 }
|
|
710
|
323
|
711 valid = spell_valid_case(mip->mi_capflags, flags);
|
300
|
712 }
|
236
|
713
|
307
|
714 if (valid)
|
300
|
715 {
|
307
|
716 if (flags & WF_BANNED)
|
|
717 res = SP_BANNED;
|
|
718 else if (flags & WF_REGION)
|
300
|
719 {
|
|
720 /* Check region. */
|
|
721 if ((mip->mi_lp->lp_region & (flags >> 8)) != 0)
|
|
722 res = SP_OK;
|
|
723 else
|
|
724 res = SP_LOCAL;
|
|
725 }
|
|
726 else if (flags & WF_RARE)
|
|
727 res = SP_RARE;
|
|
728 else
|
|
729 res = SP_OK;
|
307
|
730
|
|
731 /* Always use the longest match and the best result. */
|
|
732 if (mip->mi_result > res)
|
|
733 {
|
|
734 mip->mi_result = res;
|
|
735 mip->mi_end = mip->mi_word + wlen;
|
|
736 }
|
|
737 else if (mip->mi_result == res
|
|
738 && mip->mi_end < mip->mi_word + wlen)
|
|
739 mip->mi_end = mip->mi_word + wlen;
|
|
740
|
|
741 if (res == SP_OK)
|
|
742 break;
|
300
|
743 }
|
307
|
744 else
|
|
745 res = SP_BAD;
|
|
746
|
300
|
747 ++arridx;
|
|
748 }
|
|
749
|
307
|
750 if (res == SP_OK)
|
300
|
751 break;
|
|
752 }
|
236
|
753 }
|
|
754
|
323
|
755 /*
|
|
756 * Check case flags for a word. Return TRUE if the word has the requested
|
|
757 * case.
|
|
758 */
|
|
759 static int
|
|
760 spell_valid_case(origflags, treeflags)
|
|
761 int origflags; /* flags for the checked word. */
|
|
762 int treeflags; /* flags for the word in the spell tree */
|
|
763 {
|
|
764 return (origflags == WF_ALLCAP
|
|
765 || ((treeflags & (WF_ALLCAP | WF_KEEPCAP)) == 0
|
|
766 && ((treeflags & WF_ONECAP) == 0 || origflags == WF_ONECAP)));
|
|
767 }
|
|
768
|
300
|
769
|
236
|
770 /*
|
|
771 * Move to next spell error.
|
323
|
772 * "curline" is TRUE for "z?": find word under/after cursor in the same line.
|
236
|
773 * Return OK if found, FAIL otherwise.
|
|
774 */
|
|
775 int
|
323
|
776 spell_move_to(dir, allwords, curline)
|
236
|
777 int dir; /* FORWARD or BACKWARD */
|
|
778 int allwords; /* TRUE for "[s" and "]s" */
|
323
|
779 int curline;
|
236
|
780 {
|
249
|
781 linenr_T lnum;
|
|
782 pos_T found_pos;
|
236
|
783 char_u *line;
|
|
784 char_u *p;
|
|
785 int attr = 0;
|
|
786 int len;
|
249
|
787 int has_syntax = syntax_present(curbuf);
|
|
788 int col;
|
|
789 int can_spell;
|
236
|
790
|
310
|
791 if (!curwin->w_p_spell || *curbuf->b_p_spl == NUL)
|
236
|
792 {
|
|
793 EMSG(_("E756: Spell checking not enabled"));
|
|
794 return FAIL;
|
|
795 }
|
|
796
|
249
|
797 /*
|
|
798 * Start looking for bad word at the start of the line, because we can't
|
|
799 * start halfway a word, we don't know where it starts or ends.
|
|
800 *
|
|
801 * When searching backwards, we continue in the line to find the last
|
|
802 * bad word (in the cursor line: before the cursor).
|
|
803 */
|
|
804 lnum = curwin->w_cursor.lnum;
|
|
805 found_pos.lnum = 0;
|
236
|
806
|
|
807 while (!got_int)
|
|
808 {
|
249
|
809 line = ml_get(lnum);
|
|
810 p = line;
|
|
811
|
236
|
812 while (*p != NUL)
|
|
813 {
|
300
|
814 /* When searching backward don't search after the cursor. */
|
|
815 if (dir == BACKWARD
|
|
816 && lnum == curwin->w_cursor.lnum
|
|
817 && (colnr_T)(p - line) >= curwin->w_cursor.col)
|
|
818 break;
|
249
|
819
|
300
|
820 /* start of word */
|
|
821 len = spell_check(curwin, p, &attr);
|
249
|
822
|
300
|
823 if (attr != 0)
|
|
824 {
|
|
825 /* We found a bad word. Check the attribute. */
|
|
826 /* TODO: check for syntax @Spell cluster. */
|
|
827 if (allwords || attr == highlight_attr[HLF_SPB])
|
236
|
828 {
|
300
|
829 /* When searching forward only accept a bad word after
|
|
830 * the cursor. */
|
|
831 if (dir == BACKWARD
|
|
832 || lnum > curwin->w_cursor.lnum
|
|
833 || (lnum == curwin->w_cursor.lnum
|
323
|
834 && (colnr_T)(curline ? p - line + len
|
|
835 : p - line)
|
300
|
836 > curwin->w_cursor.col))
|
236
|
837 {
|
300
|
838 if (has_syntax)
|
249
|
839 {
|
300
|
840 col = p - line;
|
|
841 (void)syn_get_id(lnum, (colnr_T)col,
|
|
842 FALSE, &can_spell);
|
249
|
843
|
300
|
844 /* have to get the line again, a multi-line
|
|
845 * regexp may make it invalid */
|
|
846 line = ml_get(lnum);
|
|
847 p = line + col;
|
|
848 }
|
|
849 else
|
|
850 can_spell = TRUE;
|
249
|
851
|
300
|
852 if (can_spell)
|
|
853 {
|
|
854 found_pos.lnum = lnum;
|
|
855 found_pos.col = p - line;
|
249
|
856 #ifdef FEAT_VIRTUALEDIT
|
300
|
857 found_pos.coladd = 0;
|
249
|
858 #endif
|
300
|
859 if (dir == FORWARD)
|
|
860 {
|
|
861 /* No need to search further. */
|
|
862 curwin->w_cursor = found_pos;
|
|
863 return OK;
|
249
|
864 }
|
|
865 }
|
236
|
866 }
|
|
867 }
|
300
|
868 attr = 0;
|
236
|
869 }
|
|
870
|
300
|
871 /* advance to character after the word */
|
|
872 p += len;
|
|
873 if (*p == NUL)
|
|
874 break;
|
236
|
875 }
|
|
876
|
323
|
877 if (curline)
|
|
878 return FAIL; /* only check cursor line */
|
|
879
|
236
|
880 /* Advance to next line. */
|
249
|
881 if (dir == BACKWARD)
|
|
882 {
|
|
883 if (found_pos.lnum != 0)
|
|
884 {
|
|
885 /* Use the last match in the line. */
|
|
886 curwin->w_cursor = found_pos;
|
|
887 return OK;
|
|
888 }
|
|
889 if (lnum == 1)
|
|
890 return FAIL;
|
|
891 --lnum;
|
|
892 }
|
|
893 else
|
|
894 {
|
|
895 if (lnum == curbuf->b_ml.ml_line_count)
|
|
896 return FAIL;
|
|
897 ++lnum;
|
|
898 }
|
236
|
899
|
|
900 line_breakcheck();
|
|
901 }
|
|
902
|
|
903 return FAIL; /* interrupted */
|
|
904 }
|
|
905
|
|
906 /*
|
307
|
907 * Load word list(s) for "lang" from Vim spell file(s).
|
310
|
908 * "lang" must be the language without the region: e.g., "en".
|
236
|
909 */
|
307
|
910 static void
|
236
|
911 spell_load_lang(lang)
|
|
912 char_u *lang;
|
|
913 {
|
310
|
914 char_u fname_enc[85];
|
236
|
915 int r;
|
307
|
916 char_u langcp[MAXWLEN + 1];
|
|
917
|
310
|
918 /* Copy the language name to pass it to spell_load_cb() as a cookie.
|
307
|
919 * It's truncated when an error is detected. */
|
|
920 STRCPY(langcp, lang);
|
|
921
|
310
|
922 /*
|
|
923 * Find the first spell file for "lang" in 'runtimepath' and load it.
|
|
924 */
|
|
925 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
|
926 "spell/%s.%s.spl", lang, spell_enc());
|
|
927 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
307
|
928
|
|
929 if (r == FAIL && *langcp != NUL)
|
|
930 {
|
|
931 /* Try loading the ASCII version. */
|
310
|
932 vim_snprintf((char *)fname_enc, sizeof(fname_enc) - 5,
|
272
|
933 "spell/%s.ascii.spl", lang);
|
310
|
934 r = do_in_runtimepath(fname_enc, FALSE, spell_load_cb, &langcp);
|
307
|
935 }
|
|
936
|
|
937 if (r == FAIL)
|
|
938 smsg((char_u *)_("Warning: Cannot find word list \"%s\""),
|
236
|
939 fname_enc + 6);
|
310
|
940 else if (*langcp != NUL)
|
|
941 {
|
|
942 /* Load all the additions. */
|
|
943 STRCPY(fname_enc + STRLEN(fname_enc) - 3, "add.spl");
|
|
944 do_in_runtimepath(fname_enc, TRUE, spell_load_cb, &langcp);
|
|
945 }
|
|
946 }
|
|
947
|
|
948 /*
|
|
949 * Return the encoding used for spell checking: Use 'encoding', except that we
|
|
950 * use "latin1" for "latin9". And limit to 60 characters (just in case).
|
|
951 */
|
|
952 static char_u *
|
|
953 spell_enc()
|
|
954 {
|
|
955
|
|
956 #ifdef FEAT_MBYTE
|
|
957 if (STRLEN(p_enc) < 60 && STRCMP(p_enc, "iso-8859-15") != 0)
|
|
958 return p_enc;
|
|
959 #endif
|
|
960 return (char_u *)"latin1";
|
236
|
961 }
|
|
962
|
|
963 /*
|
|
964 * Allocate a new slang_T.
|
|
965 * Caller must fill "sl_next".
|
|
966 */
|
|
967 static slang_T *
|
|
968 slang_alloc(lang)
|
|
969 char_u *lang;
|
|
970 {
|
|
971 slang_T *lp;
|
|
972
|
300
|
973 lp = (slang_T *)alloc_clear(sizeof(slang_T));
|
236
|
974 if (lp != NULL)
|
|
975 {
|
|
976 lp->sl_name = vim_strsave(lang);
|
323
|
977 ga_init2(&lp->sl_rep, sizeof(fromto_T), 10);
|
|
978 ga_init2(&lp->sl_sal, sizeof(fromto_T), 10);
|
236
|
979 }
|
|
980 return lp;
|
|
981 }
|
|
982
|
|
983 /*
|
|
984 * Free the contents of an slang_T and the structure itself.
|
|
985 */
|
|
986 static void
|
|
987 slang_free(lp)
|
|
988 slang_T *lp;
|
|
989 {
|
|
990 vim_free(lp->sl_name);
|
310
|
991 vim_free(lp->sl_fname);
|
|
992 slang_clear(lp);
|
|
993 vim_free(lp);
|
|
994 }
|
|
995
|
|
996 /*
|
|
997 * Clear an slang_T so that the file can be reloaded.
|
|
998 */
|
|
999 static void
|
|
1000 slang_clear(lp)
|
|
1001 slang_T *lp;
|
|
1002 {
|
323
|
1003 garray_T *gap;
|
|
1004 fromto_T *ftp;
|
|
1005 int round;
|
|
1006
|
300
|
1007 vim_free(lp->sl_fbyts);
|
310
|
1008 lp->sl_fbyts = NULL;
|
300
|
1009 vim_free(lp->sl_kbyts);
|
310
|
1010 lp->sl_kbyts = NULL;
|
300
|
1011 vim_free(lp->sl_fidxs);
|
310
|
1012 lp->sl_fidxs = NULL;
|
300
|
1013 vim_free(lp->sl_kidxs);
|
310
|
1014 lp->sl_kidxs = NULL;
|
323
|
1015
|
|
1016 for (round = 1; round <= 2; ++round)
|
|
1017 {
|
|
1018 gap = round == 1 ? &lp->sl_rep : &lp->sl_sal;
|
|
1019 while (gap->ga_len > 0)
|
|
1020 {
|
|
1021 ftp = &((fromto_T *)gap->ga_data)[--gap->ga_len];
|
|
1022 vim_free(ftp->ft_from);
|
|
1023 vim_free(ftp->ft_to);
|
|
1024 }
|
|
1025 ga_clear(gap);
|
|
1026 }
|
|
1027
|
|
1028 vim_free(lp->sl_map);
|
|
1029 lp->sl_map = NULL;
|
236
|
1030 }
|
|
1031
|
|
1032 /*
|
307
|
1033 * Load one spell file and store the info into a slang_T.
|
236
|
1034 * Invoked through do_in_runtimepath().
|
|
1035 */
|
|
1036 static void
|
310
|
1037 spell_load_cb(fname, cookie)
|
236
|
1038 char_u *fname;
|
307
|
1039 void *cookie; /* points to the language name */
|
236
|
1040 {
|
323
|
1041 (void)spell_load_file(fname, (char_u *)cookie, NULL, FALSE);
|
310
|
1042 }
|
|
1043
|
|
1044 /*
|
|
1045 * Load one spell file and store the info into a slang_T.
|
|
1046 *
|
|
1047 * This is invoked in two ways:
|
|
1048 * - From spell_load_cb() to load a spell file for the first time. "lang" is
|
|
1049 * the language name, "old_lp" is NULL. Will allocate an slang_T.
|
|
1050 * - To reload a spell file that was changed. "lang" is NULL and "old_lp"
|
|
1051 * points to the existing slang_T.
|
323
|
1052 * Returns the slang_T the spell file was loaded into. NULL for error.
|
310
|
1053 */
|
323
|
1054 static slang_T *
|
|
1055 spell_load_file(fname, lang, old_lp, silent)
|
310
|
1056 char_u *fname;
|
|
1057 char_u *lang;
|
|
1058 slang_T *old_lp;
|
323
|
1059 int silent; /* no error if file doesn't exist */
|
310
|
1060 {
|
236
|
1061 FILE *fd;
|
|
1062 char_u buf[MAXWLEN + 1];
|
|
1063 char_u *p;
|
|
1064 int i;
|
300
|
1065 int len;
|
236
|
1066 int round;
|
|
1067 char_u *save_sourcing_name = sourcing_name;
|
|
1068 linenr_T save_sourcing_lnum = sourcing_lnum;
|
255
|
1069 int cnt, ccnt;
|
|
1070 char_u *fol;
|
307
|
1071 slang_T *lp = NULL;
|
323
|
1072 garray_T *gap;
|
|
1073 fromto_T *ftp;
|
|
1074 int rr;
|
|
1075 short *first;
|
236
|
1076
|
310
|
1077 fd = mch_fopen((char *)fname, "r");
|
236
|
1078 if (fd == NULL)
|
|
1079 {
|
323
|
1080 if (!silent)
|
|
1081 EMSG2(_(e_notopen), fname);
|
|
1082 else if (p_verbose > 2)
|
|
1083 {
|
|
1084 verbose_enter();
|
|
1085 smsg((char_u *)e_notopen, fname);
|
|
1086 verbose_leave();
|
|
1087 }
|
255
|
1088 goto endFAIL;
|
236
|
1089 }
|
310
|
1090 if (p_verbose > 2)
|
|
1091 {
|
|
1092 verbose_enter();
|
|
1093 smsg((char_u *)_("Reading spell file \"%s\""), fname);
|
|
1094 verbose_leave();
|
|
1095 }
|
|
1096
|
|
1097 if (old_lp == NULL)
|
|
1098 {
|
|
1099 lp = slang_alloc(lang);
|
|
1100 if (lp == NULL)
|
|
1101 goto endFAIL;
|
|
1102
|
|
1103 /* Remember the file name, used to reload the file when it's updated. */
|
|
1104 lp->sl_fname = vim_strsave(fname);
|
|
1105 if (lp->sl_fname == NULL)
|
|
1106 goto endFAIL;
|
|
1107
|
|
1108 /* Check for .add.spl. */
|
|
1109 lp->sl_add = strstr((char *)gettail(fname), ".add.") != NULL;
|
|
1110 }
|
|
1111 else
|
|
1112 lp = old_lp;
|
307
|
1113
|
236
|
1114 /* Set sourcing_name, so that error messages mention the file name. */
|
|
1115 sourcing_name = fname;
|
|
1116 sourcing_lnum = 0;
|
|
1117
|
255
|
1118 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
1119 * <charflagslen> <charflags> <fcharslen> <fchars> */
|
236
|
1120 for (i = 0; i < VIMSPELLMAGICL; ++i)
|
|
1121 buf[i] = getc(fd); /* <fileID> */
|
|
1122 if (STRNCMP(buf, VIMSPELLMAGIC, VIMSPELLMAGICL) != 0)
|
|
1123 {
|
|
1124 EMSG(_("E757: Wrong file ID in spell file"));
|
255
|
1125 goto endFAIL;
|
236
|
1126 }
|
|
1127
|
|
1128 cnt = getc(fd); /* <regioncnt> */
|
255
|
1129 if (cnt < 0)
|
236
|
1130 {
|
|
1131 truncerr:
|
|
1132 EMSG(_("E758: Truncated spell file"));
|
255
|
1133 goto endFAIL;
|
236
|
1134 }
|
|
1135 if (cnt > 8)
|
|
1136 {
|
|
1137 formerr:
|
307
|
1138 EMSG(_(e_format));
|
255
|
1139 goto endFAIL;
|
236
|
1140 }
|
|
1141 for (i = 0; i < cnt; ++i)
|
|
1142 {
|
|
1143 lp->sl_regions[i * 2] = getc(fd); /* <regionname> */
|
|
1144 lp->sl_regions[i * 2 + 1] = getc(fd);
|
|
1145 }
|
|
1146 lp->sl_regions[cnt * 2] = NUL;
|
|
1147
|
255
|
1148 cnt = getc(fd); /* <charflagslen> */
|
|
1149 if (cnt > 0)
|
|
1150 {
|
300
|
1151 p = alloc((unsigned)cnt);
|
255
|
1152 if (p == NULL)
|
|
1153 goto endFAIL;
|
|
1154 for (i = 0; i < cnt; ++i)
|
|
1155 p[i] = getc(fd); /* <charflags> */
|
|
1156
|
|
1157 ccnt = (getc(fd) << 8) + getc(fd); /* <fcharslen> */
|
|
1158 if (ccnt <= 0)
|
300
|
1159 {
|
|
1160 vim_free(p);
|
255
|
1161 goto formerr;
|
300
|
1162 }
|
|
1163 fol = alloc((unsigned)ccnt + 1);
|
255
|
1164 if (fol == NULL)
|
300
|
1165 {
|
|
1166 vim_free(p);
|
255
|
1167 goto endFAIL;
|
300
|
1168 }
|
255
|
1169 for (i = 0; i < ccnt; ++i)
|
|
1170 fol[i] = getc(fd); /* <fchars> */
|
|
1171 fol[i] = NUL;
|
|
1172
|
|
1173 /* Set the word-char flags and fill spell_isupper() table. */
|
300
|
1174 i = set_spell_charflags(p, cnt, fol);
|
|
1175 vim_free(p);
|
|
1176 vim_free(fol);
|
|
1177 if (i == FAIL)
|
255
|
1178 goto formerr;
|
|
1179 }
|
|
1180 else
|
|
1181 {
|
|
1182 /* When <charflagslen> is zero then <fcharlen> must also be zero. */
|
|
1183 cnt = (getc(fd) << 8) + getc(fd);
|
|
1184 if (cnt != 0)
|
|
1185 goto formerr;
|
|
1186 }
|
|
1187
|
323
|
1188 /* <SUGGEST> : <repcount> <rep> ...
|
|
1189 * <salflags> <salcount> <sal> ...
|
|
1190 * <maplen> <mapstr> */
|
|
1191 for (round = 1; round <= 2; ++round)
|
|
1192 {
|
|
1193 if (round == 1)
|
|
1194 {
|
|
1195 gap = &lp->sl_rep;
|
|
1196 first = lp->sl_rep_first;
|
|
1197 }
|
|
1198 else
|
|
1199 {
|
|
1200 gap = &lp->sl_sal;
|
|
1201 first = lp->sl_sal_first;
|
|
1202
|
|
1203 i = getc(fd); /* <salflags> */
|
|
1204 if (i & SAL_F0LLOWUP)
|
|
1205 lp->sl_followup = TRUE;
|
|
1206 if (i & SAL_COLLAPSE)
|
|
1207 lp->sl_collapse = TRUE;
|
|
1208 if (i & SAL_REM_ACCENTS)
|
|
1209 lp->sl_rem_accents = TRUE;
|
|
1210 }
|
|
1211
|
|
1212 cnt = (getc(fd) << 8) + getc(fd); /* <repcount> or <salcount> */
|
|
1213 if (cnt < 0)
|
|
1214 goto formerr;
|
|
1215
|
|
1216 if (ga_grow(gap, cnt) == FAIL)
|
|
1217 goto endFAIL;
|
|
1218 for (; gap->ga_len < cnt; ++gap->ga_len)
|
|
1219 {
|
|
1220 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
1221 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
1222 ftp = &((fromto_T *)gap->ga_data)[gap->ga_len];
|
|
1223 for (rr = 1; rr <= 2; ++rr)
|
|
1224 {
|
|
1225 ccnt = getc(fd);
|
|
1226 if (ccnt < 0)
|
|
1227 {
|
|
1228 if (rr == 2)
|
|
1229 vim_free(ftp->ft_from);
|
|
1230 goto formerr;
|
|
1231 }
|
|
1232 if ((p = alloc(ccnt + 1)) == NULL)
|
|
1233 {
|
|
1234 if (rr == 2)
|
|
1235 vim_free(ftp->ft_from);
|
|
1236 goto endFAIL;
|
|
1237 }
|
|
1238 for (i = 0; i < ccnt; ++i)
|
|
1239 p[i] = getc(fd); /* <repfrom> or <salfrom> */
|
|
1240 p[i] = NUL;
|
|
1241 if (rr == 1)
|
|
1242 ftp->ft_from = p;
|
|
1243 else
|
|
1244 ftp->ft_to = p;
|
|
1245 }
|
|
1246 }
|
|
1247
|
|
1248 /* Fill the first-index table. */
|
|
1249 for (i = 0; i < 256; ++i)
|
|
1250 first[i] = -1;
|
|
1251 for (i = 0; i < gap->ga_len; ++i)
|
|
1252 {
|
|
1253 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
1254 if (first[*ftp->ft_from] == -1)
|
|
1255 first[*ftp->ft_from] = i;
|
|
1256 }
|
|
1257 }
|
|
1258
|
|
1259 cnt = (getc(fd) << 8) + getc(fd); /* <maplen> */
|
|
1260 if (cnt < 0)
|
|
1261 goto formerr;
|
|
1262 p = alloc(cnt + 1);
|
|
1263 if (p == NULL)
|
|
1264 goto endFAIL;
|
|
1265 for (i = 0; i < cnt; ++i)
|
|
1266 p[i] = getc(fd); /* <mapstr> */
|
|
1267 p[i] = NUL;
|
|
1268 lp->sl_map = p;
|
|
1269
|
236
|
1270
|
300
|
1271 /* round 1: <LWORDTREE>
|
|
1272 * round 2: <KWORDTREE> */
|
|
1273 for (round = 1; round <= 2; ++round)
|
236
|
1274 {
|
300
|
1275 /* The tree size was computed when writing the file, so that we can
|
|
1276 * allocate it as one long block. <nodecount> */
|
|
1277 len = (getc(fd) << 24) + (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
1278 if (len < 0)
|
|
1279 goto truncerr;
|
|
1280 if (len > 0)
|
236
|
1281 {
|
300
|
1282 /* Allocate the byte array. */
|
|
1283 p = lalloc((long_u)len, TRUE);
|
|
1284 if (p == NULL)
|
|
1285 goto endFAIL;
|
|
1286 if (round == 1)
|
|
1287 lp->sl_fbyts = p;
|
|
1288 else
|
|
1289 lp->sl_kbyts = p;
|
236
|
1290
|
300
|
1291 /* Allocate the index array. */
|
|
1292 p = lalloc_clear((long_u)(len * sizeof(int)), TRUE);
|
|
1293 if (p == NULL)
|
|
1294 goto endFAIL;
|
|
1295 if (round == 1)
|
|
1296 lp->sl_fidxs = (int *)p;
|
|
1297 else
|
|
1298 lp->sl_kidxs = (int *)p;
|
|
1299
|
|
1300
|
|
1301 /* Read the tree and store it in the array. */
|
|
1302 i = read_tree(fd,
|
|
1303 round == 1 ? lp->sl_fbyts : lp->sl_kbyts,
|
|
1304 round == 1 ? lp->sl_fidxs : lp->sl_kidxs,
|
|
1305 len, 0);
|
|
1306 if (i == -1)
|
|
1307 goto truncerr;
|
|
1308 if (i < 0)
|
236
|
1309 goto formerr;
|
|
1310 }
|
300
|
1311 }
|
243
|
1312
|
310
|
1313 /* For a new file link it in the list of spell files. */
|
|
1314 if (old_lp == NULL)
|
|
1315 {
|
|
1316 lp->sl_next = first_lang;
|
|
1317 first_lang = lp;
|
|
1318 }
|
307
|
1319
|
255
|
1320 goto endOK;
|
|
1321
|
|
1322 endFAIL:
|
310
|
1323 if (lang != NULL)
|
|
1324 /* truncating the name signals the error to spell_load_lang() */
|
|
1325 *lang = NUL;
|
|
1326 if (lp != NULL && old_lp == NULL)
|
323
|
1327 {
|
307
|
1328 slang_free(lp);
|
323
|
1329 lp = NULL;
|
|
1330 }
|
255
|
1331
|
|
1332 endOK:
|
236
|
1333 if (fd != NULL)
|
|
1334 fclose(fd);
|
|
1335 sourcing_name = save_sourcing_name;
|
|
1336 sourcing_lnum = save_sourcing_lnum;
|
323
|
1337
|
|
1338 return lp;
|
236
|
1339 }
|
|
1340
|
|
1341 /*
|
300
|
1342 * Read one row of siblings from the spell file and store it in the byte array
|
|
1343 * "byts" and index array "idxs". Recursively read the children.
|
|
1344 *
|
|
1345 * NOTE: The code here must match put_tree().
|
|
1346 *
|
|
1347 * Returns the index follosing the siblings.
|
|
1348 * Returns -1 if the file is shorter than expected.
|
|
1349 * Returns -2 if there is a format error.
|
236
|
1350 */
|
300
|
1351 static int
|
|
1352 read_tree(fd, byts, idxs, maxidx, startidx)
|
|
1353 FILE *fd;
|
|
1354 char_u *byts;
|
|
1355 int *idxs;
|
|
1356 int maxidx; /* size of arrays */
|
|
1357 int startidx; /* current index in "byts" and "idxs" */
|
236
|
1358 {
|
300
|
1359 int len;
|
|
1360 int i;
|
|
1361 int n;
|
|
1362 int idx = startidx;
|
|
1363 int c;
|
|
1364 #define SHARED_MASK 0x8000000
|
236
|
1365
|
300
|
1366 len = getc(fd); /* <siblingcount> */
|
|
1367 if (len <= 0)
|
|
1368 return -1;
|
|
1369
|
|
1370 if (startidx + len >= maxidx)
|
|
1371 return -2;
|
|
1372 byts[idx++] = len;
|
|
1373
|
|
1374 /* Read the byte values, flag/region bytes and shared indexes. */
|
|
1375 for (i = 1; i <= len; ++i)
|
236
|
1376 {
|
300
|
1377 c = getc(fd); /* <byte> */
|
|
1378 if (c < 0)
|
|
1379 return -1;
|
|
1380 if (c <= BY_SPECIAL)
|
|
1381 {
|
|
1382 if (c == BY_NOFLAGS)
|
|
1383 {
|
|
1384 /* No flags, all regions. */
|
|
1385 idxs[idx] = 0;
|
|
1386 c = 0;
|
|
1387 }
|
|
1388 else if (c == BY_FLAGS)
|
|
1389 {
|
|
1390 /* Read flags and option region. */
|
|
1391 c = getc(fd); /* <flags> */
|
|
1392 if (c & WF_REGION)
|
|
1393 c = (getc(fd) << 8) + c; /* <region> */
|
|
1394 idxs[idx] = c;
|
|
1395 c = 0;
|
|
1396 }
|
|
1397 else /* c == BY_INDEX */
|
|
1398 {
|
|
1399 /* <nodeidx> */
|
|
1400 n = (getc(fd) << 16) + (getc(fd) << 8) + getc(fd);
|
|
1401 if (n < 0 || n >= maxidx)
|
|
1402 return -2;
|
|
1403 idxs[idx] = n + SHARED_MASK;
|
|
1404 c = getc(fd); /* <xbyte> */
|
|
1405 }
|
|
1406 }
|
|
1407 byts[idx++] = c;
|
236
|
1408 }
|
|
1409
|
300
|
1410 /* Recursively read the children for non-shared siblings.
|
|
1411 * Skip the end-of-word ones (zero byte value) and the shared ones (and
|
|
1412 * remove SHARED_MASK) */
|
|
1413 for (i = 1; i <= len; ++i)
|
|
1414 if (byts[startidx + i] != 0)
|
|
1415 {
|
|
1416 if (idxs[startidx + i] & SHARED_MASK)
|
|
1417 idxs[startidx + i] &= ~SHARED_MASK;
|
|
1418 else
|
|
1419 {
|
|
1420 idxs[startidx + i] = idx;
|
|
1421 idx = read_tree(fd, byts, idxs, maxidx, idx);
|
|
1422 if (idx < 0)
|
|
1423 break;
|
|
1424 }
|
|
1425 }
|
236
|
1426
|
300
|
1427 return idx;
|
236
|
1428 }
|
|
1429
|
|
1430 /*
|
|
1431 * Parse 'spelllang' and set buf->b_langp accordingly.
|
|
1432 * Returns an error message or NULL.
|
|
1433 */
|
|
1434 char_u *
|
|
1435 did_set_spelllang(buf)
|
|
1436 buf_T *buf;
|
|
1437 {
|
|
1438 garray_T ga;
|
|
1439 char_u *lang;
|
|
1440 char_u *e;
|
|
1441 char_u *region;
|
|
1442 int region_mask;
|
|
1443 slang_T *lp;
|
|
1444 int c;
|
|
1445 char_u lbuf[MAXWLEN + 1];
|
323
|
1446 char_u spf_name[MAXPATHL];
|
|
1447 int did_spf = FALSE;
|
236
|
1448
|
|
1449 ga_init2(&ga, sizeof(langp_T), 2);
|
|
1450
|
323
|
1451 /* Get the name of the .spl file associated with 'spellfile'. */
|
|
1452 if (*buf->b_p_spf == NUL)
|
|
1453 did_spf = TRUE;
|
|
1454 else
|
|
1455 vim_snprintf((char *)spf_name, sizeof(spf_name), "%s.spl",
|
|
1456 buf->b_p_spf);
|
|
1457
|
236
|
1458 /* loop over comma separated languages. */
|
|
1459 for (lang = buf->b_p_spl; *lang != NUL; lang = e)
|
|
1460 {
|
|
1461 e = vim_strchr(lang, ',');
|
|
1462 if (e == NULL)
|
|
1463 e = lang + STRLEN(lang);
|
240
|
1464 region = NULL;
|
236
|
1465 if (e > lang + 2)
|
|
1466 {
|
|
1467 if (e - lang >= MAXWLEN)
|
|
1468 {
|
|
1469 ga_clear(&ga);
|
|
1470 return e_invarg;
|
|
1471 }
|
|
1472 if (lang[2] == '_')
|
|
1473 region = lang + 3;
|
|
1474 }
|
|
1475
|
307
|
1476 /* Check if we loaded this language before. */
|
236
|
1477 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
1478 if (STRNICMP(lp->sl_name, lang, 2) == 0)
|
|
1479 break;
|
|
1480
|
|
1481 if (lp == NULL)
|
|
1482 {
|
|
1483 /* Not found, load the language. */
|
323
|
1484 vim_strncpy(lbuf, lang, e - lang);
|
236
|
1485 if (region != NULL)
|
|
1486 mch_memmove(lbuf + 2, lbuf + 5, e - lang - 4);
|
307
|
1487 spell_load_lang(lbuf);
|
236
|
1488 }
|
|
1489
|
307
|
1490 /*
|
|
1491 * Loop over the languages, there can be several files for each.
|
|
1492 */
|
|
1493 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
1494 if (STRNICMP(lp->sl_name, lang, 2) == 0)
|
236
|
1495 {
|
316
|
1496 region_mask = REGION_ALL;
|
|
1497 if (region != NULL)
|
236
|
1498 {
|
307
|
1499 /* find region in sl_regions */
|
|
1500 c = find_region(lp->sl_regions, region);
|
|
1501 if (c == REGION_ALL)
|
|
1502 {
|
316
|
1503 if (!lp->sl_add)
|
|
1504 {
|
|
1505 c = *e;
|
|
1506 *e = NUL;
|
|
1507 smsg((char_u *)_("Warning: region %s not supported"),
|
307
|
1508 lang);
|
316
|
1509 *e = c;
|
|
1510 }
|
307
|
1511 }
|
|
1512 else
|
|
1513 region_mask = 1 << c;
|
236
|
1514 }
|
307
|
1515
|
|
1516 if (ga_grow(&ga, 1) == FAIL)
|
|
1517 {
|
|
1518 ga_clear(&ga);
|
|
1519 return e_outofmem;
|
|
1520 }
|
|
1521 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
1522 LANGP_ENTRY(ga, ga.ga_len)->lp_region = region_mask;
|
|
1523 ++ga.ga_len;
|
323
|
1524
|
|
1525 /* Check if this is the 'spellfile' spell file. */
|
|
1526 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
1527 did_spf = TRUE;
|
236
|
1528 }
|
|
1529
|
|
1530 if (*e == ',')
|
|
1531 ++e;
|
|
1532 }
|
|
1533
|
323
|
1534 /*
|
|
1535 * Make sure the 'spellfile' file is loaded. It may be in 'runtimepath',
|
|
1536 * then it's probably loaded above already. Otherwise load it here.
|
|
1537 */
|
|
1538 if (!did_spf)
|
|
1539 {
|
|
1540 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
1541 if (fullpathcmp(spf_name, lp->sl_fname, FALSE) == FPC_SAME)
|
|
1542 break;
|
|
1543 if (lp == NULL)
|
|
1544 {
|
|
1545 vim_strncpy(lbuf, gettail(spf_name), 2);
|
|
1546 lp = spell_load_file(spf_name, lbuf, NULL, TRUE);
|
|
1547 }
|
|
1548 if (lp != NULL && ga_grow(&ga, 1) == OK)
|
|
1549 {
|
|
1550 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = lp;
|
|
1551 LANGP_ENTRY(ga, ga.ga_len)->lp_region = REGION_ALL;
|
|
1552 ++ga.ga_len;
|
|
1553 }
|
|
1554 }
|
|
1555
|
236
|
1556 /* Add a NULL entry to mark the end of the list. */
|
|
1557 if (ga_grow(&ga, 1) == FAIL)
|
|
1558 {
|
|
1559 ga_clear(&ga);
|
|
1560 return e_outofmem;
|
|
1561 }
|
|
1562 LANGP_ENTRY(ga, ga.ga_len)->lp_slang = NULL;
|
|
1563 ++ga.ga_len;
|
|
1564
|
|
1565 /* Everything is fine, store the new b_langp value. */
|
|
1566 ga_clear(&buf->b_langp);
|
|
1567 buf->b_langp = ga;
|
|
1568
|
|
1569 return NULL;
|
|
1570 }
|
|
1571
|
|
1572 /*
|
|
1573 * Find the region "region[2]" in "rp" (points to "sl_regions").
|
|
1574 * Each region is simply stored as the two characters of it's name.
|
|
1575 * Returns the index if found, REGION_ALL if not found.
|
|
1576 */
|
|
1577 static int
|
|
1578 find_region(rp, region)
|
|
1579 char_u *rp;
|
|
1580 char_u *region;
|
|
1581 {
|
|
1582 int i;
|
|
1583
|
|
1584 for (i = 0; ; i += 2)
|
|
1585 {
|
|
1586 if (rp[i] == NUL)
|
|
1587 return REGION_ALL;
|
|
1588 if (rp[i] == region[0] && rp[i + 1] == region[1])
|
|
1589 break;
|
|
1590 }
|
|
1591 return i / 2;
|
|
1592 }
|
|
1593
|
|
1594 /*
|
323
|
1595 * Return case type of word:
|
236
|
1596 * w word 0
|
300
|
1597 * Word WF_ONECAP
|
|
1598 * W WORD WF_ALLCAP
|
|
1599 * WoRd wOrd WF_KEEPCAP
|
236
|
1600 */
|
|
1601 static int
|
|
1602 captype(word, end)
|
|
1603 char_u *word;
|
323
|
1604 char_u *end; /* When NULL use up to NUL byte. */
|
236
|
1605 {
|
|
1606 char_u *p;
|
|
1607 int c;
|
|
1608 int firstcap;
|
|
1609 int allcap;
|
|
1610 int past_second = FALSE; /* past second word char */
|
|
1611
|
|
1612 /* find first letter */
|
307
|
1613 for (p = word; !SPELL_ISWORDP(p); mb_ptr_adv(p))
|
323
|
1614 if (end == NULL ? *p == NUL : p >= end)
|
236
|
1615 return 0; /* only non-word characters, illegal word */
|
|
1616 #ifdef FEAT_MBYTE
|
310
|
1617 if (has_mbyte)
|
|
1618 c = mb_ptr2char_adv(&p);
|
|
1619 else
|
236
|
1620 #endif
|
310
|
1621 c = *p++;
|
255
|
1622 firstcap = allcap = spell_isupper(c);
|
236
|
1623
|
|
1624 /*
|
|
1625 * Need to check all letters to find a word with mixed upper/lower.
|
|
1626 * But a word with an upper char only at start is a ONECAP.
|
|
1627 */
|
323
|
1628 for ( ; end == NULL ? *p != NUL : p < end; mb_ptr_adv(p))
|
307
|
1629 if (SPELL_ISWORDP(p))
|
236
|
1630 {
|
|
1631 #ifdef FEAT_MBYTE
|
|
1632 c = mb_ptr2char(p);
|
|
1633 #else
|
|
1634 c = *p;
|
|
1635 #endif
|
255
|
1636 if (!spell_isupper(c))
|
236
|
1637 {
|
|
1638 /* UUl -> KEEPCAP */
|
|
1639 if (past_second && allcap)
|
300
|
1640 return WF_KEEPCAP;
|
236
|
1641 allcap = FALSE;
|
|
1642 }
|
|
1643 else if (!allcap)
|
|
1644 /* UlU -> KEEPCAP */
|
300
|
1645 return WF_KEEPCAP;
|
236
|
1646 past_second = TRUE;
|
|
1647 }
|
|
1648
|
|
1649 if (allcap)
|
300
|
1650 return WF_ALLCAP;
|
236
|
1651 if (firstcap)
|
300
|
1652 return WF_ONECAP;
|
236
|
1653 return 0;
|
|
1654 }
|
|
1655
|
|
1656 # if defined(FEAT_MBYTE) || defined(PROTO)
|
|
1657 /*
|
|
1658 * Clear all spelling tables and reload them.
|
307
|
1659 * Used after 'encoding' is set and when ":mkspell" was used.
|
236
|
1660 */
|
|
1661 void
|
|
1662 spell_reload()
|
|
1663 {
|
|
1664 buf_T *buf;
|
|
1665 slang_T *lp;
|
316
|
1666 win_T *wp;
|
236
|
1667
|
307
|
1668 /* Initialize the table for SPELL_ISWORDP(). */
|
236
|
1669 init_spell_chartab();
|
|
1670
|
|
1671 /* Unload all allocated memory. */
|
|
1672 while (first_lang != NULL)
|
|
1673 {
|
|
1674 lp = first_lang;
|
|
1675 first_lang = lp->sl_next;
|
|
1676 slang_free(lp);
|
|
1677 }
|
|
1678
|
|
1679 /* Go through all buffers and handle 'spelllang'. */
|
|
1680 for (buf = firstbuf; buf != NULL; buf = buf->b_next)
|
|
1681 {
|
|
1682 ga_clear(&buf->b_langp);
|
316
|
1683
|
|
1684 /* Only load the wordlists when 'spelllang' is set and there is a
|
|
1685 * window for this buffer in which 'spell' is set. */
|
236
|
1686 if (*buf->b_p_spl != NUL)
|
316
|
1687 {
|
|
1688 FOR_ALL_WINDOWS(wp)
|
|
1689 if (wp->w_buffer == buf && wp->w_p_spell)
|
|
1690 {
|
|
1691 (void)did_set_spelllang(buf);
|
|
1692 # ifdef FEAT_WINDOWS
|
|
1693 break;
|
|
1694 # endif
|
|
1695 }
|
|
1696 }
|
236
|
1697 }
|
|
1698 }
|
|
1699 # endif
|
|
1700
|
310
|
1701 /*
|
|
1702 * Reload the spell file "fname" if it's loaded.
|
|
1703 */
|
|
1704 static void
|
323
|
1705 spell_reload_one(fname, added_word)
|
310
|
1706 char_u *fname;
|
323
|
1707 int added_word; /* invoked through "zg" */
|
310
|
1708 {
|
|
1709 slang_T *lp;
|
323
|
1710 int didit = FALSE;
|
310
|
1711
|
|
1712 for (lp = first_lang; lp != NULL; lp = lp->sl_next)
|
|
1713 if (fullpathcmp(fname, lp->sl_fname, FALSE) == FPC_SAME)
|
|
1714 {
|
|
1715 slang_clear(lp);
|
323
|
1716 (void)spell_load_file(fname, NULL, lp, FALSE);
|
310
|
1717 redraw_all_later(NOT_VALID);
|
323
|
1718 didit = TRUE;
|
310
|
1719 }
|
323
|
1720
|
|
1721 /* When "zg" was used and the file wasn't loaded yet, should redo
|
|
1722 * 'spelllang' to get it loaded. */
|
|
1723 if (added_word && !didit)
|
|
1724 did_set_spelllang(curbuf);
|
310
|
1725 }
|
|
1726
|
|
1727
|
236
|
1728 /*
|
|
1729 * Functions for ":mkspell".
|
|
1730 */
|
|
1731
|
300
|
1732 #define MAXLINELEN 500 /* Maximum length in bytes of a line in a .aff
|
236
|
1733 and .dic file. */
|
|
1734 /*
|
|
1735 * Main structure to store the contents of a ".aff" file.
|
|
1736 */
|
|
1737 typedef struct afffile_S
|
|
1738 {
|
|
1739 char_u *af_enc; /* "SET", normalized, alloc'ed string or NULL */
|
310
|
1740 int af_rar; /* RAR ID for rare word */
|
|
1741 int af_kep; /* KEP ID for keep-case word */
|
236
|
1742 hashtab_T af_pref; /* hashtable for prefixes, affheader_T */
|
|
1743 hashtab_T af_suff; /* hashtable for suffixes, affheader_T */
|
|
1744 } afffile_T;
|
|
1745
|
|
1746 typedef struct affentry_S affentry_T;
|
|
1747 /* Affix entry from ".aff" file. Used for prefixes and suffixes. */
|
|
1748 struct affentry_S
|
|
1749 {
|
|
1750 affentry_T *ae_next; /* next affix with same name/number */
|
|
1751 char_u *ae_chop; /* text to chop off basic word (can be NULL) */
|
|
1752 char_u *ae_add; /* text to add to basic word (can be NULL) */
|
|
1753 char_u *ae_cond; /* condition (NULL for ".") */
|
|
1754 regprog_T *ae_prog; /* regexp program for ae_cond or NULL */
|
300
|
1755 };
|
|
1756
|
|
1757 /* Affix header from ".aff" file. Used for af_pref and af_suff. */
|
|
1758 typedef struct affheader_S
|
|
1759 {
|
|
1760 char_u ah_key[2]; /* key for hashtable == name of affix entry */
|
|
1761 int ah_combine; /* suffix may combine with prefix */
|
|
1762 affentry_T *ah_first; /* first affix entry */
|
|
1763 } affheader_T;
|
|
1764
|
|
1765 #define HI2AH(hi) ((affheader_T *)(hi)->hi_key)
|
|
1766
|
|
1767 /*
|
|
1768 * Structure that is used to store the items in the word tree. This avoids
|
|
1769 * the need to keep track of each allocated thing, it's freed all at once
|
|
1770 * after ":mkspell" is done.
|
|
1771 */
|
|
1772 #define SBLOCKSIZE 16000 /* size of sb_data */
|
|
1773 typedef struct sblock_S sblock_T;
|
|
1774 struct sblock_S
|
|
1775 {
|
|
1776 sblock_T *sb_next; /* next block in list */
|
|
1777 int sb_used; /* nr of bytes already in use */
|
|
1778 char_u sb_data[1]; /* data, actually longer */
|
236
|
1779 };
|
|
1780
|
|
1781 /*
|
300
|
1782 * A node in the tree.
|
236
|
1783 */
|
300
|
1784 typedef struct wordnode_S wordnode_T;
|
|
1785 struct wordnode_S
|
236
|
1786 {
|
300
|
1787 char_u wn_hashkey[6]; /* room for the hash key */
|
|
1788 wordnode_T *wn_next; /* next node with same hash key */
|
|
1789 wordnode_T *wn_child; /* child (next byte in word) */
|
|
1790 wordnode_T *wn_sibling; /* next sibling (alternate byte in word,
|
|
1791 always sorted) */
|
|
1792 wordnode_T *wn_wnode; /* parent node that will write this node */
|
|
1793 int wn_index; /* index in written nodes (valid after first
|
|
1794 round) */
|
|
1795 char_u wn_byte; /* Byte for this node. NUL for word end */
|
|
1796 char_u wn_flags; /* when wn_byte is NUL: WF_ flags */
|
|
1797 char_u wn_region; /* when wn_byte is NUL: region mask */
|
236
|
1798 };
|
|
1799
|
300
|
1800 #define HI2WN(hi) (wordnode_T *)((hi)->hi_key)
|
236
|
1801
|
300
|
1802 /*
|
|
1803 * Info used while reading the spell files.
|
|
1804 */
|
|
1805 typedef struct spellinfo_S
|
249
|
1806 {
|
300
|
1807 wordnode_T *si_foldroot; /* tree with case-folded words */
|
|
1808 wordnode_T *si_keeproot; /* tree with keep-case words */
|
|
1809 sblock_T *si_blocks; /* memory blocks used */
|
|
1810 int si_ascii; /* handling only ASCII words */
|
310
|
1811 int si_add; /* addition file */
|
300
|
1812 int si_region; /* region mask */
|
|
1813 vimconv_T si_conv; /* for conversion to 'encoding' */
|
302
|
1814 int si_memtot; /* runtime memory used */
|
310
|
1815 int si_verbose; /* verbose messages */
|
316
|
1816 int si_region_count; /* number of regions supported (1 when there
|
|
1817 are no regions) */
|
|
1818 char_u si_region_name[16]; /* region names (if count > 1) */
|
323
|
1819
|
|
1820 garray_T si_rep; /* list of fromto_T entries from REP lines */
|
|
1821 garray_T si_sal; /* list of fromto_T entries from SAL lines */
|
|
1822 int si_followup; /* soundsalike: ? */
|
|
1823 int si_collapse; /* soundsalike: ? */
|
|
1824 int si_rem_accents; /* soundsalike: remove accents */
|
|
1825 garray_T si_map; /* MAP info concatenated */
|
300
|
1826 } spellinfo_T;
|
249
|
1827
|
300
|
1828 static afffile_T *spell_read_aff __ARGS((char_u *fname, spellinfo_T *spin));
|
323
|
1829 static void add_fromto __ARGS((spellinfo_T *spin, garray_T *gap, char_u *from, char_u *to));
|
|
1830 static int sal_to_bool __ARGS((char_u *s));
|
240
|
1831 static int has_non_ascii __ARGS((char_u *s));
|
300
|
1832 static void spell_free_aff __ARGS((afffile_T *aff));
|
|
1833 static int spell_read_dic __ARGS((char_u *fname, spellinfo_T *spin, afffile_T *affile));
|
307
|
1834 static int store_aff_word __ARGS((char_u *word, spellinfo_T *spin, char_u *afflist, hashtab_T *ht, hashtab_T *xht, int comb, int flags));
|
300
|
1835 static int spell_read_wordfile __ARGS((char_u *fname, spellinfo_T *spin));
|
|
1836 static void *getroom __ARGS((sblock_T **blp, size_t len));
|
|
1837 static char_u *getroom_save __ARGS((sblock_T **blp, char_u *s));
|
|
1838 static void free_blocks __ARGS((sblock_T *bl));
|
|
1839 static wordnode_T *wordtree_alloc __ARGS((sblock_T **blp));
|
316
|
1840 static int store_word __ARGS((char_u *word, spellinfo_T *spin, int flags, int region));
|
300
|
1841 static int tree_add_word __ARGS((char_u *word, wordnode_T *tree, int flags, int region, sblock_T **blp));
|
310
|
1842 static void wordtree_compress __ARGS((wordnode_T *root, spellinfo_T *spin));
|
300
|
1843 static int node_compress __ARGS((wordnode_T *node, hashtab_T *ht, int *tot));
|
|
1844 static int node_equal __ARGS((wordnode_T *n1, wordnode_T *n2));
|
316
|
1845 static void write_vim_spell __ARGS((char_u *fname, spellinfo_T *spin));
|
300
|
1846 static int put_tree __ARGS((FILE *fd, wordnode_T *node, int index, int regionmask));
|
323
|
1847 static void mkspell __ARGS((int fcount, char_u **fnames, int ascii, int overwrite, int added_word));
|
310
|
1848 static void init_spellfile __ARGS((void));
|
236
|
1849
|
|
1850 /*
|
323
|
1851 * Read the affix file "fname".
|
316
|
1852 * Returns an afffile_T, NULL for complete failure.
|
236
|
1853 */
|
|
1854 static afffile_T *
|
300
|
1855 spell_read_aff(fname, spin)
|
236
|
1856 char_u *fname;
|
300
|
1857 spellinfo_T *spin;
|
236
|
1858 {
|
|
1859 FILE *fd;
|
|
1860 afffile_T *aff;
|
|
1861 char_u rline[MAXLINELEN];
|
|
1862 char_u *line;
|
|
1863 char_u *pc = NULL;
|
|
1864 char_u *(items[6]);
|
|
1865 int itemcnt;
|
|
1866 char_u *p;
|
|
1867 int lnum = 0;
|
|
1868 affheader_T *cur_aff = NULL;
|
|
1869 int aff_todo = 0;
|
|
1870 hashtab_T *tp;
|
255
|
1871 char_u *low = NULL;
|
|
1872 char_u *fol = NULL;
|
|
1873 char_u *upp = NULL;
|
307
|
1874 static char *e_affname = N_("Affix name too long in %s line %d: %s");
|
323
|
1875 int do_rep;
|
|
1876 int do_sal;
|
|
1877 int do_map;
|
|
1878 int found_map = FALSE;
|
236
|
1879
|
300
|
1880 /*
|
|
1881 * Open the file.
|
|
1882 */
|
310
|
1883 fd = mch_fopen((char *)fname, "r");
|
236
|
1884 if (fd == NULL)
|
|
1885 {
|
|
1886 EMSG2(_(e_notopen), fname);
|
|
1887 return NULL;
|
|
1888 }
|
|
1889
|
310
|
1890 if (spin->si_verbose || p_verbose > 2)
|
|
1891 {
|
|
1892 if (!spin->si_verbose)
|
|
1893 verbose_enter();
|
|
1894 smsg((char_u *)_("Reading affix file %s..."), fname);
|
|
1895 out_flush();
|
|
1896 if (!spin->si_verbose)
|
|
1897 verbose_leave();
|
|
1898 }
|
236
|
1899
|
323
|
1900 /* Only do REP lines when not done in another .aff file already. */
|
|
1901 do_rep = spin->si_rep.ga_len == 0;
|
|
1902
|
|
1903 /* Only do SAL lines when not done in another .aff file already. */
|
|
1904 do_sal = spin->si_sal.ga_len == 0;
|
|
1905
|
|
1906 /* Only do MAP lines when not done in another .aff file already. */
|
|
1907 do_map = spin->si_map.ga_len == 0;
|
|
1908
|
300
|
1909 /*
|
|
1910 * Allocate and init the afffile_T structure.
|
|
1911 */
|
|
1912 aff = (afffile_T *)getroom(&spin->si_blocks, sizeof(afffile_T));
|
236
|
1913 if (aff == NULL)
|
|
1914 return NULL;
|
|
1915 hash_init(&aff->af_pref);
|
|
1916 hash_init(&aff->af_suff);
|
|
1917
|
|
1918 /*
|
|
1919 * Read all the lines in the file one by one.
|
|
1920 */
|
255
|
1921 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
236
|
1922 {
|
255
|
1923 line_breakcheck();
|
236
|
1924 ++lnum;
|
|
1925
|
|
1926 /* Skip comment lines. */
|
|
1927 if (*rline == '#')
|
|
1928 continue;
|
|
1929
|
|
1930 /* Convert from "SET" to 'encoding' when needed. */
|
|
1931 vim_free(pc);
|
310
|
1932 #ifdef FEAT_MBYTE
|
300
|
1933 if (spin->si_conv.vc_type != CONV_NONE)
|
236
|
1934 {
|
300
|
1935 pc = string_convert(&spin->si_conv, rline, NULL);
|
255
|
1936 if (pc == NULL)
|
|
1937 {
|
|
1938 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
1939 fname, lnum, rline);
|
|
1940 continue;
|
|
1941 }
|
236
|
1942 line = pc;
|
|
1943 }
|
|
1944 else
|
310
|
1945 #endif
|
236
|
1946 {
|
|
1947 pc = NULL;
|
|
1948 line = rline;
|
|
1949 }
|
|
1950
|
|
1951 /* Split the line up in white separated items. Put a NUL after each
|
|
1952 * item. */
|
|
1953 itemcnt = 0;
|
|
1954 for (p = line; ; )
|
|
1955 {
|
|
1956 while (*p != NUL && *p <= ' ') /* skip white space and CR/NL */
|
|
1957 ++p;
|
|
1958 if (*p == NUL)
|
|
1959 break;
|
300
|
1960 if (itemcnt == 6) /* too many items */
|
|
1961 break;
|
236
|
1962 items[itemcnt++] = p;
|
300
|
1963 while (*p > ' ') /* skip until white space or CR/NL */
|
236
|
1964 ++p;
|
|
1965 if (*p == NUL)
|
|
1966 break;
|
|
1967 *p++ = NUL;
|
|
1968 }
|
|
1969
|
|
1970 /* Handle non-empty lines. */
|
|
1971 if (itemcnt > 0)
|
|
1972 {
|
|
1973 if (STRCMP(items[0], "SET") == 0 && itemcnt == 2
|
|
1974 && aff->af_enc == NULL)
|
|
1975 {
|
310
|
1976 #ifdef FEAT_MBYTE
|
300
|
1977 /* Setup for conversion from "ENC" to 'encoding'. */
|
|
1978 aff->af_enc = enc_canonize(items[1]);
|
|
1979 if (aff->af_enc != NULL && !spin->si_ascii
|
|
1980 && convert_setup(&spin->si_conv, aff->af_enc,
|
|
1981 p_enc) == FAIL)
|
|
1982 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
|
1983 fname, aff->af_enc, p_enc);
|
310
|
1984 #else
|
|
1985 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
1986 #endif
|
236
|
1987 }
|
302
|
1988 else if (STRCMP(items[0], "NOSPLITSUGS") == 0 && itemcnt == 1)
|
|
1989 {
|
323
|
1990 /* ignored, we always split */
|
302
|
1991 }
|
323
|
1992 else if (STRCMP(items[0], "TRY") == 0 && itemcnt == 2)
|
300
|
1993 {
|
323
|
1994 /* ignored, we look in the tree for what chars may appear */
|
300
|
1995 }
|
307
|
1996 else if (STRCMP(items[0], "RAR") == 0 && itemcnt == 2
|
|
1997 && aff->af_rar == 0)
|
|
1998 {
|
|
1999 aff->af_rar = items[1][0];
|
|
2000 if (items[1][1] != NUL)
|
|
2001 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
2002 }
|
310
|
2003 else if (STRCMP(items[0], "KEP") == 0 && itemcnt == 2
|
|
2004 && aff->af_kep == 0)
|
307
|
2005 {
|
310
|
2006 aff->af_kep = items[1][0];
|
307
|
2007 if (items[1][1] != NUL)
|
|
2008 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
|
2009 }
|
236
|
2010 else if ((STRCMP(items[0], "PFX") == 0
|
|
2011 || STRCMP(items[0], "SFX") == 0)
|
|
2012 && aff_todo == 0
|
|
2013 && itemcnt == 4)
|
|
2014 {
|
|
2015 /* New affix letter. */
|
300
|
2016 cur_aff = (affheader_T *)getroom(&spin->si_blocks,
|
|
2017 sizeof(affheader_T));
|
236
|
2018 if (cur_aff == NULL)
|
|
2019 break;
|
|
2020 cur_aff->ah_key[0] = *items[1];
|
|
2021 cur_aff->ah_key[1] = NUL;
|
|
2022 if (items[1][1] != NUL)
|
307
|
2023 smsg((char_u *)_(e_affname), fname, lnum, items[1]);
|
236
|
2024 if (*items[2] == 'Y')
|
|
2025 cur_aff->ah_combine = TRUE;
|
300
|
2026 else if (*items[2] != 'N')
|
236
|
2027 smsg((char_u *)_("Expected Y or N in %s line %d: %s"),
|
|
2028 fname, lnum, items[2]);
|
|
2029 if (*items[0] == 'P')
|
|
2030 tp = &aff->af_pref;
|
|
2031 else
|
|
2032 tp = &aff->af_suff;
|
300
|
2033 aff_todo = atoi((char *)items[3]);
|
236
|
2034 if (!HASHITEM_EMPTY(hash_find(tp, cur_aff->ah_key)))
|
300
|
2035 {
|
236
|
2036 smsg((char_u *)_("Duplicate affix in %s line %d: %s"),
|
|
2037 fname, lnum, items[1]);
|
300
|
2038 aff_todo = 0;
|
|
2039 }
|
236
|
2040 else
|
|
2041 hash_add(tp, cur_aff->ah_key);
|
|
2042 }
|
|
2043 else if ((STRCMP(items[0], "PFX") == 0
|
|
2044 || STRCMP(items[0], "SFX") == 0)
|
|
2045 && aff_todo > 0
|
|
2046 && STRCMP(cur_aff->ah_key, items[1]) == 0
|
|
2047 && itemcnt == 5)
|
|
2048 {
|
|
2049 affentry_T *aff_entry;
|
|
2050
|
|
2051 /* New item for an affix letter. */
|
|
2052 --aff_todo;
|
300
|
2053 aff_entry = (affentry_T *)getroom(&spin->si_blocks,
|
|
2054 sizeof(affentry_T));
|
236
|
2055 if (aff_entry == NULL)
|
|
2056 break;
|
240
|
2057
|
236
|
2058 if (STRCMP(items[2], "0") != 0)
|
300
|
2059 aff_entry->ae_chop = getroom_save(&spin->si_blocks,
|
|
2060 items[2]);
|
236
|
2061 if (STRCMP(items[3], "0") != 0)
|
300
|
2062 aff_entry->ae_add = getroom_save(&spin->si_blocks,
|
|
2063 items[3]);
|
236
|
2064
|
300
|
2065 /* Don't use an affix entry with non-ASCII characters when
|
|
2066 * "spin->si_ascii" is TRUE. */
|
|
2067 if (!spin->si_ascii || !(has_non_ascii(aff_entry->ae_chop)
|
240
|
2068 || has_non_ascii(aff_entry->ae_add)))
|
|
2069 {
|
|
2070 aff_entry->ae_next = cur_aff->ah_first;
|
|
2071 cur_aff->ah_first = aff_entry;
|
300
|
2072
|
|
2073 if (STRCMP(items[4], ".") != 0)
|
|
2074 {
|
|
2075 char_u buf[MAXLINELEN];
|
|
2076
|
|
2077 aff_entry->ae_cond = getroom_save(&spin->si_blocks,
|
|
2078 items[4]);
|
|
2079 if (*items[0] == 'P')
|
|
2080 sprintf((char *)buf, "^%s", items[4]);
|
|
2081 else
|
|
2082 sprintf((char *)buf, "%s$", items[4]);
|
|
2083 aff_entry->ae_prog = vim_regcomp(buf,
|
|
2084 RE_MAGIC + RE_STRING);
|
|
2085 }
|
240
|
2086 }
|
236
|
2087 }
|
255
|
2088 else if (STRCMP(items[0], "FOL") == 0 && itemcnt == 2)
|
|
2089 {
|
|
2090 if (fol != NULL)
|
|
2091 smsg((char_u *)_("Duplicate FOL in %s line %d"),
|
|
2092 fname, lnum);
|
|
2093 else
|
|
2094 fol = vim_strsave(items[1]);
|
|
2095 }
|
|
2096 else if (STRCMP(items[0], "LOW") == 0 && itemcnt == 2)
|
|
2097 {
|
|
2098 if (low != NULL)
|
|
2099 smsg((char_u *)_("Duplicate LOW in %s line %d"),
|
|
2100 fname, lnum);
|
|
2101 else
|
|
2102 low = vim_strsave(items[1]);
|
|
2103 }
|
|
2104 else if (STRCMP(items[0], "UPP") == 0 && itemcnt == 2)
|
|
2105 {
|
|
2106 if (upp != NULL)
|
|
2107 smsg((char_u *)_("Duplicate UPP in %s line %d"),
|
|
2108 fname, lnum);
|
|
2109 else
|
|
2110 upp = vim_strsave(items[1]);
|
|
2111 }
|
236
|
2112 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 2)
|
323
|
2113 {
|
236
|
2114 /* Ignore REP count */;
|
323
|
2115 if (!isdigit(*items[1]))
|
|
2116 smsg((char_u *)_("Expected REP count in %s line %d"),
|
|
2117 fname, lnum);
|
|
2118 }
|
236
|
2119 else if (STRCMP(items[0], "REP") == 0 && itemcnt == 3)
|
|
2120 {
|
|
2121 /* REP item */
|
323
|
2122 if (do_rep)
|
|
2123 add_fromto(spin, &spin->si_rep, items[1], items[2]);
|
|
2124 }
|
|
2125 else if (STRCMP(items[0], "MAP") == 0 && itemcnt == 2)
|
|
2126 {
|
|
2127 /* MAP item or count */
|
|
2128 if (!found_map)
|
|
2129 {
|
|
2130 /* First line contains the count. */
|
|
2131 found_map = TRUE;
|
|
2132 if (!isdigit(*items[1]))
|
|
2133 smsg((char_u *)_("Expected MAP count in %s line %d"),
|
|
2134 fname, lnum);
|
|
2135 }
|
|
2136 else if (do_map)
|
|
2137 {
|
|
2138 /* We simply concatenate all the MAP strings, separated by
|
|
2139 * slashes. */
|
|
2140 ga_concat(&spin->si_map, items[1]);
|
|
2141 ga_append(&spin->si_map, '/');
|
|
2142 }
|
|
2143 }
|
|
2144 else if (STRCMP(items[0], "SAL") == 0 && itemcnt == 3)
|
|
2145 {
|
|
2146 if (do_sal)
|
|
2147 {
|
|
2148 /* SAL item (sounds-a-like)
|
|
2149 * Either one of the known keys or a from-to pair. */
|
|
2150 if (STRCMP(items[1], "followup") == 0)
|
|
2151 spin->si_followup = sal_to_bool(items[2]);
|
|
2152 else if (STRCMP(items[1], "collapse_result") == 0)
|
|
2153 spin->si_collapse = sal_to_bool(items[2]);
|
|
2154 else if (STRCMP(items[1], "remove_accents") == 0)
|
|
2155 spin->si_rem_accents = sal_to_bool(items[2]);
|
|
2156 else
|
|
2157 /* when "to" is "_" it means empty */
|
|
2158 add_fromto(spin, &spin->si_sal, items[1],
|
|
2159 STRCMP(items[2], "_") == 0 ? (char_u *)""
|
|
2160 : items[2]);
|
|
2161 }
|
236
|
2162 }
|
300
|
2163 else
|
236
|
2164 smsg((char_u *)_("Unrecognized item in %s line %d: %s"),
|
|
2165 fname, lnum, items[0]);
|
|
2166 }
|
|
2167 }
|
|
2168
|
255
|
2169 if (fol != NULL || low != NULL || upp != NULL)
|
|
2170 {
|
316
|
2171 /*
|
|
2172 * Don't write a word table for an ASCII file, so that we don't check
|
|
2173 * for conflicts with a word table that matches 'encoding'.
|
|
2174 * Don't write one for utf-8 either, we use utf_isupper() and
|
|
2175 * mb_get_class(), the list of chars in the file will be incomplete.
|
|
2176 */
|
|
2177 if (!spin->si_ascii
|
|
2178 #ifdef FEAT_MBYTE
|
|
2179 && !enc_utf8
|
|
2180 #endif
|
|
2181 )
|
260
|
2182 {
|
|
2183 if (fol == NULL || low == NULL || upp == NULL)
|
|
2184 smsg((char_u *)_("Missing FOL/LOW/UPP line in %s"), fname);
|
|
2185 else
|
316
|
2186 (void)set_spell_chartab(fol, low, upp);
|
260
|
2187 }
|
255
|
2188
|
|
2189 vim_free(fol);
|
|
2190 vim_free(low);
|
|
2191 vim_free(upp);
|
|
2192 }
|
|
2193
|
236
|
2194 vim_free(pc);
|
|
2195 fclose(fd);
|
|
2196 return aff;
|
|
2197 }
|
|
2198
|
|
2199 /*
|
323
|
2200 * Add a from-to item to "gap". Used for REP and SAL items.
|
|
2201 * They are stored case-folded.
|
|
2202 */
|
|
2203 static void
|
|
2204 add_fromto(spin, gap, from, to)
|
|
2205 spellinfo_T *spin;
|
|
2206 garray_T *gap;
|
|
2207 char_u *from;
|
|
2208 char_u *to;
|
|
2209 {
|
|
2210 fromto_T *ftp;
|
|
2211 char_u word[MAXWLEN];
|
|
2212
|
|
2213 if (ga_grow(gap, 1) == OK)
|
|
2214 {
|
|
2215 ftp = ((fromto_T *)gap->ga_data) + gap->ga_len;
|
|
2216 (void)spell_casefold(from, STRLEN(from), word, MAXWLEN);
|
|
2217 ftp->ft_from = getroom_save(&spin->si_blocks, word);
|
|
2218 (void)spell_casefold(to, STRLEN(to), word, MAXWLEN);
|
|
2219 ftp->ft_to = getroom_save(&spin->si_blocks, word);
|
|
2220 ++gap->ga_len;
|
|
2221 }
|
|
2222 }
|
|
2223
|
|
2224 /*
|
|
2225 * Convert a boolean argument in a SAL line to TRUE or FALSE;
|
|
2226 */
|
|
2227 static int
|
|
2228 sal_to_bool(s)
|
|
2229 char_u *s;
|
|
2230 {
|
|
2231 return STRCMP(s, "1") == 0 || STRCMP(s, "true") == 0;
|
|
2232 }
|
|
2233
|
|
2234 /*
|
240
|
2235 * Return TRUE if string "s" contains a non-ASCII character (128 or higher).
|
|
2236 * When "s" is NULL FALSE is returned.
|
|
2237 */
|
|
2238 static int
|
|
2239 has_non_ascii(s)
|
|
2240 char_u *s;
|
|
2241 {
|
|
2242 char_u *p;
|
|
2243
|
|
2244 if (s != NULL)
|
|
2245 for (p = s; *p != NUL; ++p)
|
|
2246 if (*p >= 128)
|
|
2247 return TRUE;
|
|
2248 return FALSE;
|
|
2249 }
|
|
2250
|
|
2251 /*
|
236
|
2252 * Free the structure filled by spell_read_aff().
|
|
2253 */
|
|
2254 static void
|
|
2255 spell_free_aff(aff)
|
|
2256 afffile_T *aff;
|
|
2257 {
|
|
2258 hashtab_T *ht;
|
|
2259 hashitem_T *hi;
|
|
2260 int todo;
|
|
2261 affheader_T *ah;
|
300
|
2262 affentry_T *ae;
|
236
|
2263
|
|
2264 vim_free(aff->af_enc);
|
|
2265
|
300
|
2266 /* All this trouble to foree the "ae_prog" items... */
|
236
|
2267 for (ht = &aff->af_pref; ; ht = &aff->af_suff)
|
|
2268 {
|
|
2269 todo = ht->ht_used;
|
|
2270 for (hi = ht->ht_array; todo > 0; ++hi)
|
|
2271 {
|
|
2272 if (!HASHITEM_EMPTY(hi))
|
|
2273 {
|
|
2274 --todo;
|
|
2275 ah = HI2AH(hi);
|
300
|
2276 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
|
2277 vim_free(ae->ae_prog);
|
236
|
2278 }
|
|
2279 }
|
|
2280 if (ht == &aff->af_suff)
|
|
2281 break;
|
|
2282 }
|
300
|
2283
|
236
|
2284 hash_clear(&aff->af_pref);
|
|
2285 hash_clear(&aff->af_suff);
|
|
2286 }
|
|
2287
|
|
2288 /*
|
300
|
2289 * Read dictionary file "fname".
|
236
|
2290 * Returns OK or FAIL;
|
|
2291 */
|
|
2292 static int
|
300
|
2293 spell_read_dic(fname, spin, affile)
|
236
|
2294 char_u *fname;
|
300
|
2295 spellinfo_T *spin;
|
|
2296 afffile_T *affile;
|
236
|
2297 {
|
300
|
2298 hashtab_T ht;
|
236
|
2299 char_u line[MAXLINELEN];
|
300
|
2300 char_u *afflist;
|
|
2301 char_u *dw;
|
236
|
2302 char_u *pc;
|
|
2303 char_u *w;
|
|
2304 int l;
|
|
2305 hash_T hash;
|
|
2306 hashitem_T *hi;
|
|
2307 FILE *fd;
|
|
2308 int lnum = 1;
|
300
|
2309 int non_ascii = 0;
|
|
2310 int retval = OK;
|
|
2311 char_u message[MAXLINELEN + MAXWLEN];
|
307
|
2312 int flags;
|
236
|
2313
|
300
|
2314 /*
|
|
2315 * Open the file.
|
|
2316 */
|
310
|
2317 fd = mch_fopen((char *)fname, "r");
|
236
|
2318 if (fd == NULL)
|
|
2319 {
|
|
2320 EMSG2(_(e_notopen), fname);
|
|
2321 return FAIL;
|
|
2322 }
|
|
2323
|
300
|
2324 /* The hashtable is only used to detect duplicated words. */
|
|
2325 hash_init(&ht);
|
|
2326
|
310
|
2327 if (spin->si_verbose || p_verbose > 2)
|
|
2328 {
|
|
2329 if (!spin->si_verbose)
|
|
2330 verbose_enter();
|
|
2331 smsg((char_u *)_("Reading dictionary file %s..."), fname);
|
|
2332 out_flush();
|
|
2333 if (!spin->si_verbose)
|
|
2334 verbose_leave();
|
|
2335 }
|
236
|
2336
|
|
2337 /* Read and ignore the first line: word count. */
|
|
2338 (void)vim_fgets(line, MAXLINELEN, fd);
|
|
2339 if (!isdigit(*skipwhite(line)))
|
|
2340 EMSG2(_("E760: No word count in %s"), fname);
|
|
2341
|
|
2342 /*
|
|
2343 * Read all the lines in the file one by one.
|
|
2344 * The words are converted to 'encoding' here, before being added to
|
|
2345 * the hashtable.
|
|
2346 */
|
255
|
2347 while (!vim_fgets(line, MAXLINELEN, fd) && !got_int)
|
236
|
2348 {
|
255
|
2349 line_breakcheck();
|
236
|
2350 ++lnum;
|
|
2351
|
300
|
2352 /* Remove CR, LF and white space from the end. White space halfway
|
|
2353 * the word is kept to allow e.g., "et al.". */
|
236
|
2354 l = STRLEN(line);
|
|
2355 while (l > 0 && line[l - 1] <= ' ')
|
|
2356 --l;
|
|
2357 if (l == 0)
|
|
2358 continue; /* empty line */
|
|
2359 line[l] = NUL;
|
|
2360
|
300
|
2361 /* This takes time, print a message now and then. */
|
310
|
2362 if (spin->si_verbose && (lnum & 0x3ff) == 0)
|
300
|
2363 {
|
|
2364 vim_snprintf((char *)message, sizeof(message),
|
|
2365 _("line %6d - %s"), lnum, line);
|
|
2366 msg_start();
|
|
2367 msg_outtrans_attr(message, 0);
|
|
2368 msg_clr_eos();
|
|
2369 msg_didout = FALSE;
|
|
2370 msg_col = 0;
|
|
2371 out_flush();
|
|
2372 }
|
|
2373
|
236
|
2374 /* Find the optional affix names. */
|
300
|
2375 afflist = vim_strchr(line, '/');
|
|
2376 if (afflist != NULL)
|
|
2377 *afflist++ = NUL;
|
236
|
2378
|
300
|
2379 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
2380 if (spin->si_ascii && has_non_ascii(line))
|
|
2381 {
|
|
2382 ++non_ascii;
|
240
|
2383 continue;
|
300
|
2384 }
|
240
|
2385
|
310
|
2386 #ifdef FEAT_MBYTE
|
236
|
2387 /* Convert from "SET" to 'encoding' when needed. */
|
300
|
2388 if (spin->si_conv.vc_type != CONV_NONE)
|
236
|
2389 {
|
300
|
2390 pc = string_convert(&spin->si_conv, line, NULL);
|
255
|
2391 if (pc == NULL)
|
|
2392 {
|
|
2393 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
2394 fname, lnum, line);
|
|
2395 continue;
|
|
2396 }
|
236
|
2397 w = pc;
|
|
2398 }
|
|
2399 else
|
310
|
2400 #endif
|
236
|
2401 {
|
|
2402 pc = NULL;
|
|
2403 w = line;
|
|
2404 }
|
|
2405
|
300
|
2406 /* Store the word in the hashtable to be able to find duplicates. */
|
|
2407 dw = (char_u *)getroom_save(&spin->si_blocks, w);
|
236
|
2408 if (dw == NULL)
|
300
|
2409 retval = FAIL;
|
|
2410 vim_free(pc);
|
|
2411 if (retval == FAIL)
|
236
|
2412 break;
|
|
2413
|
300
|
2414 hash = hash_hash(dw);
|
|
2415 hi = hash_lookup(&ht, dw, hash);
|
236
|
2416 if (!HASHITEM_EMPTY(hi))
|
|
2417 smsg((char_u *)_("Duplicate word in %s line %d: %s"),
|
300
|
2418 fname, lnum, line);
|
236
|
2419 else
|
300
|
2420 hash_add_item(&ht, hi, dw, hash);
|
|
2421
|
307
|
2422 flags = 0;
|
|
2423 if (afflist != NULL)
|
|
2424 {
|
|
2425 /* Check for affix name that stands for keep-case word and stands
|
|
2426 * for rare word (if defined). */
|
310
|
2427 if (affile->af_kep != NUL
|
|
2428 && vim_strchr(afflist, affile->af_kep) != NULL)
|
307
|
2429 flags |= WF_KEEPCAP;
|
|
2430 if (affile->af_rar != NUL
|
|
2431 && vim_strchr(afflist, affile->af_rar) != NULL)
|
|
2432 flags |= WF_RARE;
|
|
2433 }
|
|
2434
|
300
|
2435 /* Add the word to the word tree(s). */
|
316
|
2436 if (store_word(dw, spin, flags, spin->si_region) == FAIL)
|
300
|
2437 retval = FAIL;
|
236
|
2438
|
300
|
2439 if (afflist != NULL)
|
|
2440 {
|
|
2441 /* Find all matching suffixes and add the resulting words.
|
|
2442 * Additionally do matching prefixes that combine. */
|
|
2443 if (store_aff_word(dw, spin, afflist,
|
307
|
2444 &affile->af_suff, &affile->af_pref,
|
|
2445 FALSE, flags) == FAIL)
|
300
|
2446 retval = FAIL;
|
|
2447
|
|
2448 /* Find all matching prefixes and add the resulting words. */
|
|
2449 if (store_aff_word(dw, spin, afflist,
|
307
|
2450 &affile->af_pref, NULL, FALSE, flags) == FAIL)
|
300
|
2451 retval = FAIL;
|
|
2452 }
|
236
|
2453 }
|
|
2454
|
300
|
2455 if (spin->si_ascii && non_ascii > 0)
|
|
2456 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
2457 non_ascii);
|
|
2458 hash_clear(&ht);
|
|
2459
|
236
|
2460 fclose(fd);
|
300
|
2461 return retval;
|
236
|
2462 }
|
|
2463
|
|
2464 /*
|
300
|
2465 * Apply affixes to a word and store the resulting words.
|
|
2466 * "ht" is the hashtable with affentry_T that need to be applied, either
|
|
2467 * prefixes or suffixes.
|
|
2468 * "xht", when not NULL, is the prefix hashtable, to be used additionally on
|
|
2469 * the resulting words for combining affixes.
|
|
2470 *
|
|
2471 * Returns FAIL when out of memory.
|
236
|
2472 */
|
300
|
2473 static int
|
307
|
2474 store_aff_word(word, spin, afflist, ht, xht, comb, flags)
|
300
|
2475 char_u *word; /* basic word start */
|
|
2476 spellinfo_T *spin; /* spell info */
|
|
2477 char_u *afflist; /* list of names of supported affixes */
|
|
2478 hashtab_T *ht;
|
|
2479 hashtab_T *xht;
|
|
2480 int comb; /* only use affixes that combine */
|
307
|
2481 int flags; /* flags for the word */
|
236
|
2482 {
|
|
2483 int todo;
|
|
2484 hashitem_T *hi;
|
300
|
2485 affheader_T *ah;
|
|
2486 affentry_T *ae;
|
|
2487 regmatch_T regmatch;
|
|
2488 char_u newword[MAXWLEN];
|
|
2489 int retval = OK;
|
|
2490 int i;
|
|
2491 char_u *p;
|
236
|
2492
|
300
|
2493 todo = ht->ht_used;
|
|
2494 for (hi = ht->ht_array; todo > 0 && retval == OK; ++hi)
|
236
|
2495 {
|
|
2496 if (!HASHITEM_EMPTY(hi))
|
|
2497 {
|
|
2498 --todo;
|
300
|
2499 ah = HI2AH(hi);
|
236
|
2500
|
300
|
2501 /* Check that the affix combines, if required, and that the word
|
|
2502 * supports this affix. */
|
|
2503 if ((!comb || ah->ah_combine)
|
|
2504 && vim_strchr(afflist, *ah->ah_key) != NULL)
|
236
|
2505 {
|
300
|
2506 /* Loop over all affix entries with this name. */
|
|
2507 for (ae = ah->ah_first; ae != NULL; ae = ae->ae_next)
|
236
|
2508 {
|
300
|
2509 /* Check the condition. It's not logical to match case
|
|
2510 * here, but it is required for compatibility with
|
|
2511 * Myspell. */
|
|
2512 regmatch.regprog = ae->ae_prog;
|
|
2513 regmatch.rm_ic = FALSE;
|
|
2514 if (ae->ae_prog == NULL
|
|
2515 || vim_regexec(®match, word, (colnr_T)0))
|
|
2516 {
|
|
2517 /* Match. Remove the chop and add the affix. */
|
|
2518 if (xht == NULL)
|
240
|
2519 {
|
300
|
2520 /* prefix: chop/add at the start of the word */
|
|
2521 if (ae->ae_add == NULL)
|
|
2522 *newword = NUL;
|
|
2523 else
|
|
2524 STRCPY(newword, ae->ae_add);
|
|
2525 p = word;
|
|
2526 if (ae->ae_chop != NULL)
|
310
|
2527 {
|
300
|
2528 /* Skip chop string. */
|
310
|
2529 #ifdef FEAT_MBYTE
|
|
2530 if (has_mbyte)
|
|
2531 i = mb_charlen(ae->ae_chop);
|
|
2532 else
|
|
2533 #endif
|
|
2534 i = STRLEN(ae->ae_chop);
|
|
2535 for ( ; i > 0; --i)
|
300
|
2536 mb_ptr_adv(p);
|
310
|
2537 }
|
300
|
2538 STRCAT(newword, p);
|
|
2539 }
|
|
2540 else
|
|
2541 {
|
|
2542 /* suffix: chop/add at the end of the word */
|
|
2543 STRCPY(newword, word);
|
|
2544 if (ae->ae_chop != NULL)
|
|
2545 {
|
|
2546 /* Remove chop string. */
|
|
2547 p = newword + STRLEN(newword);
|
310
|
2548 #ifdef FEAT_MBYTE
|
|
2549 if (has_mbyte)
|
|
2550 i = mb_charlen(ae->ae_chop);
|
|
2551 else
|
|
2552 #endif
|
|
2553 i = STRLEN(ae->ae_chop);
|
|
2554 for ( ; i > 0; --i)
|
300
|
2555 mb_ptr_back(newword, p);
|
|
2556 *p = NUL;
|
|
2557 }
|
|
2558 if (ae->ae_add != NULL)
|
|
2559 STRCAT(newword, ae->ae_add);
|
240
|
2560 }
|
|
2561
|
300
|
2562 /* Store the modified word. */
|
316
|
2563 if (store_word(newword, spin,
|
|
2564 flags, spin->si_region) == FAIL)
|
300
|
2565 retval = FAIL;
|
236
|
2566
|
300
|
2567 /* When added a suffix and combining is allowed also
|
|
2568 * try adding prefixes additionally. */
|
|
2569 if (xht != NULL && ah->ah_combine)
|
|
2570 if (store_aff_word(newword, spin, afflist,
|
307
|
2571 xht, NULL, TRUE, flags) == FAIL)
|
300
|
2572 retval = FAIL;
|
236
|
2573 }
|
|
2574 }
|
|
2575 }
|
|
2576 }
|
|
2577 }
|
|
2578
|
|
2579 return retval;
|
|
2580 }
|
|
2581
|
|
2582 /*
|
300
|
2583 * Read a file with a list of words.
|
236
|
2584 */
|
|
2585 static int
|
300
|
2586 spell_read_wordfile(fname, spin)
|
|
2587 char_u *fname;
|
|
2588 spellinfo_T *spin;
|
236
|
2589 {
|
300
|
2590 FILE *fd;
|
|
2591 long lnum = 0;
|
|
2592 char_u rline[MAXLINELEN];
|
|
2593 char_u *line;
|
|
2594 char_u *pc = NULL;
|
|
2595 int l;
|
|
2596 int retval = OK;
|
|
2597 int did_word = FALSE;
|
|
2598 int non_ascii = 0;
|
307
|
2599 int flags;
|
316
|
2600 int regionmask;
|
236
|
2601
|
300
|
2602 /*
|
|
2603 * Open the file.
|
|
2604 */
|
310
|
2605 fd = mch_fopen((char *)fname, "r");
|
300
|
2606 if (fd == NULL)
|
236
|
2607 {
|
300
|
2608 EMSG2(_(e_notopen), fname);
|
|
2609 return FAIL;
|
236
|
2610 }
|
|
2611
|
310
|
2612 if (spin->si_verbose || p_verbose > 2)
|
|
2613 {
|
|
2614 if (!spin->si_verbose)
|
|
2615 verbose_enter();
|
|
2616 smsg((char_u *)_("Reading word file %s..."), fname);
|
|
2617 out_flush();
|
|
2618 if (!spin->si_verbose)
|
|
2619 verbose_leave();
|
|
2620 }
|
300
|
2621
|
|
2622 /*
|
|
2623 * Read all the lines in the file one by one.
|
|
2624 */
|
|
2625 while (!vim_fgets(rline, MAXLINELEN, fd) && !got_int)
|
|
2626 {
|
|
2627 line_breakcheck();
|
|
2628 ++lnum;
|
|
2629
|
|
2630 /* Skip comment lines. */
|
|
2631 if (*rline == '#')
|
|
2632 continue;
|
|
2633
|
|
2634 /* Remove CR, LF and white space from the end. */
|
|
2635 l = STRLEN(rline);
|
|
2636 while (l > 0 && rline[l - 1] <= ' ')
|
|
2637 --l;
|
|
2638 if (l == 0)
|
|
2639 continue; /* empty or blank line */
|
|
2640 rline[l] = NUL;
|
|
2641
|
|
2642 /* Convert from "=encoding={encoding}" to 'encoding' when needed. */
|
|
2643 vim_free(pc);
|
310
|
2644 #ifdef FEAT_MBYTE
|
300
|
2645 if (spin->si_conv.vc_type != CONV_NONE)
|
|
2646 {
|
|
2647 pc = string_convert(&spin->si_conv, rline, NULL);
|
|
2648 if (pc == NULL)
|
|
2649 {
|
|
2650 smsg((char_u *)_("Conversion failure for word in %s line %d: %s"),
|
|
2651 fname, lnum, rline);
|
|
2652 continue;
|
|
2653 }
|
|
2654 line = pc;
|
|
2655 }
|
|
2656 else
|
310
|
2657 #endif
|
300
|
2658 {
|
|
2659 pc = NULL;
|
|
2660 line = rline;
|
|
2661 }
|
|
2662
|
307
|
2663 flags = 0;
|
316
|
2664 regionmask = spin->si_region;
|
307
|
2665
|
|
2666 if (*line == '/')
|
300
|
2667 {
|
307
|
2668 ++line;
|
316
|
2669
|
307
|
2670 if (STRNCMP(line, "encoding=", 9) == 0)
|
300
|
2671 {
|
|
2672 if (spin->si_conv.vc_type != CONV_NONE)
|
316
|
2673 smsg((char_u *)_("Duplicate /encoding= line ignored in %s line %d: %s"),
|
|
2674 fname, lnum, line - 1);
|
300
|
2675 else if (did_word)
|
316
|
2676 smsg((char_u *)_("/encoding= line after word ignored in %s line %d: %s"),
|
|
2677 fname, lnum, line - 1);
|
300
|
2678 else
|
|
2679 {
|
310
|
2680 #ifdef FEAT_MBYTE
|
|
2681 char_u *enc;
|
|
2682
|
300
|
2683 /* Setup for conversion to 'encoding'. */
|
316
|
2684 line += 10;
|
|
2685 enc = enc_canonize(line);
|
300
|
2686 if (enc != NULL && !spin->si_ascii
|
|
2687 && convert_setup(&spin->si_conv, enc,
|
|
2688 p_enc) == FAIL)
|
|
2689 smsg((char_u *)_("Conversion in %s not supported: from %s to %s"),
|
316
|
2690 fname, line, p_enc);
|
300
|
2691 vim_free(enc);
|
310
|
2692 #else
|
|
2693 smsg((char_u *)_("Conversion in %s not supported"), fname);
|
|
2694 #endif
|
300
|
2695 }
|
307
|
2696 continue;
|
300
|
2697 }
|
307
|
2698
|
316
|
2699 if (STRNCMP(line, "regions=", 8) == 0)
|
|
2700 {
|
|
2701 if (spin->si_region_count > 1)
|
|
2702 smsg((char_u *)_("Duplicate /regions= line ignored in %s line %d: %s"),
|
|
2703 fname, lnum, line);
|
|
2704 else
|
|
2705 {
|
|
2706 line += 8;
|
|
2707 if (STRLEN(line) > 16)
|
|
2708 smsg((char_u *)_("Too many regions in %s line %d: %s"),
|
|
2709 fname, lnum, line);
|
|
2710 else
|
|
2711 {
|
|
2712 spin->si_region_count = STRLEN(line) / 2;
|
|
2713 STRCPY(spin->si_region_name, line);
|
|
2714 }
|
|
2715 }
|
|
2716 continue;
|
|
2717 }
|
|
2718
|
307
|
2719 if (*line == '=')
|
|
2720 {
|
|
2721 /* keep-case word */
|
|
2722 flags |= WF_KEEPCAP;
|
|
2723 ++line;
|
|
2724 }
|
|
2725
|
|
2726 if (*line == '!')
|
|
2727 {
|
|
2728 /* Bad, bad, wicked word. */
|
|
2729 flags |= WF_BANNED;
|
|
2730 ++line;
|
|
2731 }
|
|
2732 else if (*line == '?')
|
|
2733 {
|
|
2734 /* Rare word. */
|
|
2735 flags |= WF_RARE;
|
|
2736 ++line;
|
|
2737 }
|
|
2738
|
316
|
2739 if (VIM_ISDIGIT(*line))
|
|
2740 {
|
|
2741 /* region number(s) */
|
|
2742 regionmask = 0;
|
|
2743 while (VIM_ISDIGIT(*line))
|
|
2744 {
|
|
2745 l = *line - '0';
|
|
2746 if (l > spin->si_region_count)
|
|
2747 {
|
|
2748 smsg((char_u *)_("Invalid region nr in %s line %d: %s"),
|
|
2749 fname, lnum, line);
|
|
2750 break;
|
|
2751 }
|
|
2752 regionmask |= 1 << (l - 1);
|
|
2753 ++line;
|
|
2754 }
|
|
2755 flags |= WF_REGION;
|
|
2756 }
|
|
2757
|
307
|
2758 if (flags == 0)
|
|
2759 {
|
|
2760 smsg((char_u *)_("/ line ignored in %s line %d: %s"),
|
300
|
2761 fname, lnum, line);
|
307
|
2762 continue;
|
|
2763 }
|
300
|
2764 }
|
|
2765
|
|
2766 /* Skip non-ASCII words when "spin->si_ascii" is TRUE. */
|
|
2767 if (spin->si_ascii && has_non_ascii(line))
|
|
2768 {
|
|
2769 ++non_ascii;
|
|
2770 continue;
|
|
2771 }
|
|
2772
|
|
2773 /* Normal word: store it. */
|
316
|
2774 if (store_word(line, spin, flags, regionmask) == FAIL)
|
300
|
2775 {
|
|
2776 retval = FAIL;
|
|
2777 break;
|
|
2778 }
|
|
2779 did_word = TRUE;
|
|
2780 }
|
|
2781
|
|
2782 vim_free(pc);
|
|
2783 fclose(fd);
|
|
2784
|
310
|
2785 if (spin->si_ascii && non_ascii > 0 && (spin->si_verbose || p_verbose > 2))
|
|
2786 {
|
|
2787 if (p_verbose > 2)
|
|
2788 verbose_enter();
|
300
|
2789 smsg((char_u *)_("Ignored %d words with non-ASCII characters"),
|
|
2790 non_ascii);
|
310
|
2791 if (p_verbose > 2)
|
|
2792 verbose_leave();
|
|
2793 }
|
300
|
2794 return retval;
|
236
|
2795 }
|
|
2796
|
|
2797 /*
|
300
|
2798 * Get part of an sblock_T, "len" bytes long.
|
|
2799 * This avoids calling free() for every little struct we use.
|
|
2800 * The memory is cleared to all zeros.
|
|
2801 * Returns NULL when out of memory.
|
|
2802 */
|
|
2803 static void *
|
|
2804 getroom(blp, len)
|
|
2805 sblock_T **blp;
|
|
2806 size_t len; /* length needed */
|
|
2807 {
|
|
2808 char_u *p;
|
|
2809 sblock_T *bl = *blp;
|
|
2810
|
|
2811 if (bl == NULL || bl->sb_used + len > SBLOCKSIZE)
|
|
2812 {
|
|
2813 /* Allocate a block of memory. This is not freed until much later. */
|
|
2814 bl = (sblock_T *)alloc_clear((unsigned)(sizeof(sblock_T) + SBLOCKSIZE));
|
|
2815 if (bl == NULL)
|
|
2816 return NULL;
|
|
2817 bl->sb_next = *blp;
|
|
2818 *blp = bl;
|
|
2819 bl->sb_used = 0;
|
|
2820 }
|
|
2821
|
|
2822 p = bl->sb_data + bl->sb_used;
|
|
2823 bl->sb_used += len;
|
|
2824
|
|
2825 return p;
|
|
2826 }
|
|
2827
|
|
2828 /*
|
|
2829 * Make a copy of a string into memory allocated with getroom().
|
|
2830 */
|
|
2831 static char_u *
|
|
2832 getroom_save(blp, s)
|
|
2833 sblock_T **blp;
|
|
2834 char_u *s;
|
|
2835 {
|
|
2836 char_u *sc;
|
|
2837
|
|
2838 sc = (char_u *)getroom(blp, STRLEN(s) + 1);
|
|
2839 if (sc != NULL)
|
|
2840 STRCPY(sc, s);
|
|
2841 return sc;
|
|
2842 }
|
|
2843
|
|
2844
|
|
2845 /*
|
|
2846 * Free the list of allocated sblock_T.
|
236
|
2847 */
|
|
2848 static void
|
300
|
2849 free_blocks(bl)
|
|
2850 sblock_T *bl;
|
236
|
2851 {
|
300
|
2852 sblock_T *next;
|
236
|
2853
|
300
|
2854 while (bl != NULL)
|
236
|
2855 {
|
300
|
2856 next = bl->sb_next;
|
|
2857 vim_free(bl);
|
|
2858 bl = next;
|
236
|
2859 }
|
|
2860 }
|
|
2861
|
|
2862 /*
|
300
|
2863 * Allocate the root of a word tree.
|
236
|
2864 */
|
300
|
2865 static wordnode_T *
|
|
2866 wordtree_alloc(blp)
|
|
2867 sblock_T **blp;
|
236
|
2868 {
|
300
|
2869 return (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
236
|
2870 }
|
|
2871
|
|
2872 /*
|
300
|
2873 * Store a word in the tree(s).
|
307
|
2874 * Always store it in the case-folded tree. A keep-case word can also be used
|
|
2875 * with all caps.
|
300
|
2876 * For a keep-case word also store it in the keep-case tree.
|
236
|
2877 */
|
|
2878 static int
|
316
|
2879 store_word(word, spin, flags, region)
|
300
|
2880 char_u *word;
|
|
2881 spellinfo_T *spin;
|
307
|
2882 int flags; /* extra flags, WF_BANNED */
|
316
|
2883 int region; /* supported region(s) */
|
236
|
2884 {
|
300
|
2885 int len = STRLEN(word);
|
|
2886 int ct = captype(word, word + len);
|
|
2887 char_u foldword[MAXWLEN];
|
|
2888 int res;
|
236
|
2889
|
323
|
2890 (void)spell_casefold(word, len, foldword, MAXWLEN);
|
|
2891 res = tree_add_word(foldword, spin->si_foldroot, ct | flags,
|
|
2892 region, &spin->si_blocks);
|
307
|
2893
|
|
2894 if (res == OK && (ct == WF_KEEPCAP || flags & WF_KEEPCAP))
|
|
2895 res = tree_add_word(word, spin->si_keeproot, flags,
|
316
|
2896 region, &spin->si_blocks);
|
300
|
2897 return res;
|
236
|
2898 }
|
|
2899
|
|
2900 /*
|
300
|
2901 * Add word "word" to a word tree at "root".
|
255
|
2902 * Returns FAIL when out of memory.
|
236
|
2903 */
|
255
|
2904 static int
|
300
|
2905 tree_add_word(word, root, flags, region, blp)
|
|
2906 char_u *word;
|
|
2907 wordnode_T *root;
|
|
2908 int flags;
|
|
2909 int region;
|
|
2910 sblock_T **blp;
|
236
|
2911 {
|
300
|
2912 wordnode_T *node = root;
|
|
2913 wordnode_T *np;
|
|
2914 wordnode_T **prev = NULL;
|
|
2915 int i;
|
255
|
2916
|
300
|
2917 /* Add each byte of the word to the tree, including the NUL at the end. */
|
|
2918 for (i = 0; ; ++i)
|
255
|
2919 {
|
300
|
2920 /* Look for the sibling that has the same character. They are sorted
|
|
2921 * on byte value, thus stop searching when a sibling is found with a
|
|
2922 * higher byte value. For zero bytes (end of word) check that the
|
|
2923 * flags are equal, there is a separate zero byte for each flag value.
|
|
2924 */
|
|
2925 while (node != NULL && (node->wn_byte < word[i]
|
307
|
2926 || (node->wn_byte == 0 && node->wn_flags != (flags & 0xff))))
|
236
|
2927 {
|
300
|
2928 prev = &node->wn_sibling;
|
|
2929 node = *prev;
|
236
|
2930 }
|
300
|
2931 if (node == NULL || node->wn_byte != word[i])
|
255
|
2932 {
|
300
|
2933 /* Allocate a new node. */
|
|
2934 np = (wordnode_T *)getroom(blp, sizeof(wordnode_T));
|
|
2935 if (np == NULL)
|
|
2936 return FAIL;
|
|
2937 np->wn_byte = word[i];
|
|
2938 *prev = np;
|
|
2939 np->wn_sibling = node;
|
|
2940 node = np;
|
255
|
2941 }
|
300
|
2942
|
|
2943 if (word[i] == NUL)
|
|
2944 {
|
|
2945 node->wn_flags = flags;
|
|
2946 node->wn_region |= region;
|
|
2947 break;
|
|
2948 }
|
|
2949 prev = &node->wn_child;
|
|
2950 node = *prev;
|
255
|
2951 }
|
|
2952
|
|
2953 return OK;
|
236
|
2954 }
|
|
2955
|
|
2956 /*
|
300
|
2957 * Compress a tree: find tails that are identical and can be shared.
|
|
2958 */
|
|
2959 static void
|
310
|
2960 wordtree_compress(root, spin)
|
300
|
2961 wordnode_T *root;
|
310
|
2962 spellinfo_T *spin;
|
300
|
2963 {
|
|
2964 hashtab_T ht;
|
|
2965 int n;
|
|
2966 int tot = 0;
|
|
2967
|
|
2968 if (root != NULL)
|
|
2969 {
|
|
2970 hash_init(&ht);
|
|
2971 n = node_compress(root, &ht, &tot);
|
310
|
2972 if (spin->si_verbose || p_verbose > 2)
|
|
2973 {
|
|
2974 if (!spin->si_verbose)
|
|
2975 verbose_enter();
|
|
2976 smsg((char_u *)_("Compressed %d of %d nodes; %d%% remaining"),
|
300
|
2977 n, tot, (tot - n) * 100 / tot);
|
310
|
2978 if (p_verbose > 2)
|
|
2979 verbose_leave();
|
|
2980 }
|
300
|
2981 hash_clear(&ht);
|
|
2982 }
|
|
2983 }
|
|
2984
|
|
2985 /*
|
|
2986 * Compress a node, its siblings and its children, depth first.
|
|
2987 * Returns the number of compressed nodes.
|
236
|
2988 */
|
255
|
2989 static int
|
300
|
2990 node_compress(node, ht, tot)
|
|
2991 wordnode_T *node;
|
|
2992 hashtab_T *ht;
|
|
2993 int *tot; /* total count of nodes before compressing,
|
|
2994 incremented while going through the tree */
|
236
|
2995 {
|
300
|
2996 wordnode_T *np;
|
|
2997 wordnode_T *tp;
|
|
2998 wordnode_T *child;
|
|
2999 hash_T hash;
|
236
|
3000 hashitem_T *hi;
|
300
|
3001 int len = 0;
|
|
3002 unsigned nr, n;
|
|
3003 int compressed = 0;
|
236
|
3004
|
300
|
3005 /*
|
|
3006 * Go through the list of siblings. Compress each child and then try
|
|
3007 * finding an identical child to replace it.
|
|
3008 * Note that with "child" we mean not just the node that is pointed to,
|
|
3009 * but the whole list of siblings, of which the node is the first.
|
|
3010 */
|
|
3011 for (np = node; np != NULL; np = np->wn_sibling)
|
236
|
3012 {
|
300
|
3013 ++len;
|
|
3014 if ((child = np->wn_child) != NULL)
|
|
3015 {
|
|
3016 /* Compress the child. This fills wn_hashkey. */
|
|
3017 compressed += node_compress(child, ht, tot);
|
|
3018
|
|
3019 /* Try to find an identical child. */
|
|
3020 hash = hash_hash(child->wn_hashkey);
|
|
3021 hi = hash_lookup(ht, child->wn_hashkey, hash);
|
|
3022 tp = NULL;
|
|
3023 if (!HASHITEM_EMPTY(hi))
|
|
3024 {
|
|
3025 /* There are children with an identical hash value. Now check
|
|
3026 * if there is one that is really identical. */
|
|
3027 for (tp = HI2WN(hi); tp != NULL; tp = tp->wn_next)
|
|
3028 if (node_equal(child, tp))
|
|
3029 {
|
|
3030 /* Found one! Now use that child in place of the
|
|
3031 * current one. This means the current child is
|
|
3032 * dropped from the tree. */
|
|
3033 np->wn_child = tp;
|
|
3034 ++compressed;
|
|
3035 break;
|
|
3036 }
|
|
3037 if (tp == NULL)
|
|
3038 {
|
|
3039 /* No other child with this hash value equals the child of
|
|
3040 * the node, add it to the linked list after the first
|
|
3041 * item. */
|
|
3042 tp = HI2WN(hi);
|
|
3043 child->wn_next = tp->wn_next;
|
|
3044 tp->wn_next = child;
|
|
3045 }
|
|
3046 }
|
|
3047 else
|
|
3048 /* No other child has this hash value, add it to the
|
|
3049 * hashtable. */
|
|
3050 hash_add_item(ht, hi, child->wn_hashkey, hash);
|
|
3051 }
|
236
|
3052 }
|
300
|
3053 *tot += len;
|
|
3054
|
|
3055 /*
|
|
3056 * Make a hash key for the node and its siblings, so that we can quickly
|
|
3057 * find a lookalike node. This must be done after compressing the sibling
|
|
3058 * list, otherwise the hash key would become invalid by the compression.
|
|
3059 */
|
|
3060 node->wn_hashkey[0] = len;
|
|
3061 nr = 0;
|
|
3062 for (np = node; np != NULL; np = np->wn_sibling)
|
236
|
3063 {
|
300
|
3064 if (np->wn_byte == NUL)
|
|
3065 /* end node: only use wn_flags and wn_region */
|
|
3066 n = np->wn_flags + (np->wn_region << 8);
|
|
3067 else
|
|
3068 /* byte node: use the byte value and the child pointer */
|
|
3069 n = np->wn_byte + ((long_u)np->wn_child << 8);
|
|
3070 nr = nr * 101 + n;
|
236
|
3071 }
|
300
|
3072
|
|
3073 /* Avoid NUL bytes, it terminates the hash key. */
|
|
3074 n = nr & 0xff;
|
|
3075 node->wn_hashkey[1] = n == 0 ? 1 : n;
|
|
3076 n = (nr >> 8) & 0xff;
|
|
3077 node->wn_hashkey[2] = n == 0 ? 1 : n;
|
|
3078 n = (nr >> 16) & 0xff;
|
|
3079 node->wn_hashkey[3] = n == 0 ? 1 : n;
|
|
3080 n = (nr >> 24) & 0xff;
|
|
3081 node->wn_hashkey[4] = n == 0 ? 1 : n;
|
|
3082 node->wn_hashkey[5] = NUL;
|
|
3083
|
|
3084 return compressed;
|
|
3085 }
|
|
3086
|
|
3087 /*
|
|
3088 * Return TRUE when two nodes have identical siblings and children.
|
|
3089 */
|
|
3090 static int
|
|
3091 node_equal(n1, n2)
|
|
3092 wordnode_T *n1;
|
|
3093 wordnode_T *n2;
|
|
3094 {
|
|
3095 wordnode_T *p1;
|
|
3096 wordnode_T *p2;
|
|
3097
|
|
3098 for (p1 = n1, p2 = n2; p1 != NULL && p2 != NULL;
|
|
3099 p1 = p1->wn_sibling, p2 = p2->wn_sibling)
|
|
3100 if (p1->wn_byte != p2->wn_byte
|
|
3101 || (p1->wn_byte == NUL
|
|
3102 ? (p1->wn_flags != p2->wn_flags
|
|
3103 || p1->wn_region != p2->wn_region)
|
|
3104 : (p1->wn_child != p2->wn_child)))
|
|
3105 break;
|
|
3106
|
|
3107 return p1 == NULL && p2 == NULL;
|
236
|
3108 }
|
|
3109
|
|
3110 /*
|
|
3111 * Write a number to file "fd", MSB first, in "len" bytes.
|
|
3112 */
|
255
|
3113 void
|
236
|
3114 put_bytes(fd, nr, len)
|
|
3115 FILE *fd;
|
|
3116 long_u nr;
|
|
3117 int len;
|
|
3118 {
|
|
3119 int i;
|
|
3120
|
|
3121 for (i = len - 1; i >= 0; --i)
|
|
3122 putc((int)(nr >> (i * 8)), fd);
|
|
3123 }
|
|
3124
|
323
|
3125 static int
|
|
3126 #ifdef __BORLANDC__
|
|
3127 _RTLENTRYF
|
|
3128 #endif
|
|
3129 rep_compare __ARGS((const void *s1, const void *s2));
|
|
3130
|
|
3131 /*
|
|
3132 * Function given to qsort() to sort the REP items on "from" string.
|
|
3133 */
|
|
3134 static int
|
|
3135 #ifdef __BORLANDC__
|
|
3136 _RTLENTRYF
|
|
3137 #endif
|
|
3138 rep_compare(s1, s2)
|
|
3139 const void *s1;
|
|
3140 const void *s2;
|
|
3141 {
|
|
3142 fromto_T *p1 = (fromto_T *)s1;
|
|
3143 fromto_T *p2 = (fromto_T *)s2;
|
|
3144
|
|
3145 return STRCMP(p1->ft_from, p2->ft_from);
|
|
3146 }
|
|
3147
|
236
|
3148 /*
|
|
3149 * Write the Vim spell file "fname".
|
|
3150 */
|
|
3151 static void
|
316
|
3152 write_vim_spell(fname, spin)
|
236
|
3153 char_u *fname;
|
300
|
3154 spellinfo_T *spin;
|
236
|
3155 {
|
300
|
3156 FILE *fd;
|
|
3157 int regionmask;
|
236
|
3158 int round;
|
300
|
3159 wordnode_T *tree;
|
|
3160 int nodecount;
|
323
|
3161 int i;
|
|
3162 int l;
|
|
3163 garray_T *gap;
|
|
3164 fromto_T *ftp;
|
|
3165 char_u *p;
|
|
3166 int rr;
|
236
|
3167
|
310
|
3168 fd = mch_fopen((char *)fname, "w");
|
300
|
3169 if (fd == NULL)
|
236
|
3170 {
|
|
3171 EMSG2(_(e_notopen), fname);
|
|
3172 return;
|
|
3173 }
|
|
3174
|
255
|
3175 /* <HEADER>: <fileID> <regioncnt> <regionname> ...
|
|
3176 * <charflagslen> <charflags> <fcharslen> <fchars> */
|
300
|
3177
|
|
3178 /* <fileID> */
|
|
3179 if (fwrite(VIMSPELLMAGIC, VIMSPELLMAGICL, (size_t)1, fd) != 1)
|
|
3180 EMSG(_(e_write));
|
236
|
3181
|
|
3182 /* write the region names if there is more than one */
|
316
|
3183 if (spin->si_region_count > 1)
|
236
|
3184 {
|
316
|
3185 putc(spin->si_region_count, fd); /* <regioncnt> <regionname> ... */
|
|
3186 fwrite(spin->si_region_name, (size_t)(spin->si_region_count * 2),
|
|
3187 (size_t)1, fd);
|
|
3188 regionmask = (1 << spin->si_region_count) - 1;
|
236
|
3189 }
|
|
3190 else
|
|
3191 {
|
300
|
3192 putc(0, fd);
|
|
3193 regionmask = 0;
|
236
|
3194 }
|
|
3195
|
323
|
3196 /*
|
|
3197 * Write the table with character flags and table for case folding.
|
260
|
3198 * <charflagslen> <charflags> <fcharlen> <fchars>
|
|
3199 * Skip this for ASCII, the table may conflict with the one used for
|
323
|
3200 * 'encoding'.
|
|
3201 * Also skip this for an .add.spl file, the main spell file must contain
|
|
3202 * the table (avoids that it conflicts). File is shorter too.
|
|
3203 */
|
|
3204 if (spin->si_ascii || spin->si_add)
|
260
|
3205 {
|
300
|
3206 putc(0, fd);
|
|
3207 putc(0, fd);
|
|
3208 putc(0, fd);
|
260
|
3209 }
|
|
3210 else
|
300
|
3211 write_spell_chartab(fd);
|
255
|
3212
|
323
|
3213 /* Sort the REP items. */
|
|
3214 qsort(spin->si_rep.ga_data, (size_t)spin->si_rep.ga_len,
|
|
3215 sizeof(fromto_T), rep_compare);
|
|
3216
|
|
3217 /* <SUGGEST> : <repcount> <rep> ...
|
|
3218 * <salflags> <salcount> <sal> ...
|
|
3219 * <maplen> <mapstr> */
|
|
3220 for (round = 1; round <= 2; ++round)
|
|
3221 {
|
|
3222 if (round == 1)
|
|
3223 gap = &spin->si_rep;
|
|
3224 else
|
|
3225 {
|
|
3226 gap = &spin->si_sal;
|
|
3227
|
|
3228 i = 0;
|
|
3229 if (spin->si_followup)
|
|
3230 i |= SAL_F0LLOWUP;
|
|
3231 if (spin->si_collapse)
|
|
3232 i |= SAL_COLLAPSE;
|
|
3233 if (spin->si_rem_accents)
|
|
3234 i |= SAL_REM_ACCENTS;
|
|
3235 putc(i, fd); /* <salflags> */
|
|
3236 }
|
|
3237
|
|
3238 put_bytes(fd, (long_u)gap->ga_len, 2); /* <repcount> or <salcount> */
|
|
3239 for (i = 0; i < gap->ga_len; ++i)
|
|
3240 {
|
|
3241 /* <rep> : <repfromlen> <repfrom> <reptolen> <repto> */
|
|
3242 /* <sal> : <salfromlen> <salfrom> <saltolen> <salto> */
|
|
3243 ftp = &((fromto_T *)gap->ga_data)[i];
|
|
3244 for (rr = 1; rr <= 2; ++rr)
|
|
3245 {
|
|
3246 p = rr == 1 ? ftp->ft_from : ftp->ft_to;
|
|
3247 l = STRLEN(p);
|
|
3248 putc(l, fd);
|
|
3249 fwrite(p, l, (size_t)1, fd);
|
|
3250 }
|
|
3251 }
|
|
3252 }
|
|
3253
|
|
3254 put_bytes(fd, (long_u)spin->si_map.ga_len, 2); /* <maplen> */
|
|
3255 if (spin->si_map.ga_len > 0) /* <mapstr> */
|
|
3256 fwrite(spin->si_map.ga_data, (size_t)spin->si_map.ga_len,
|
|
3257 (size_t)1, fd);
|
302
|
3258
|
236
|
3259 /*
|
300
|
3260 * <LWORDTREE> <KWORDTREE>
|
236
|
3261 */
|
323
|
3262 spin->si_memtot = 0;
|
300
|
3263 for (round = 1; round <= 2; ++round)
|
236
|
3264 {
|
300
|
3265 tree = (round == 1) ? spin->si_foldroot : spin->si_keeproot;
|
236
|
3266
|
300
|
3267 /* Count the number of nodes. Needed to be able to allocate the
|
|
3268 * memory when reading the nodes. Also fills in the index for shared
|
|
3269 * nodes. */
|
|
3270 nodecount = put_tree(NULL, tree, 0, regionmask);
|
236
|
3271
|
300
|
3272 /* number of nodes in 4 bytes */
|
|
3273 put_bytes(fd, (long_u)nodecount, 4); /* <nodecount> */
|
302
|
3274 spin->si_memtot += nodecount + nodecount * sizeof(int);
|
236
|
3275
|
300
|
3276 /* Write the nodes. */
|
|
3277 (void)put_tree(fd, tree, 0, regionmask);
|
236
|
3278 }
|
|
3279
|
300
|
3280 fclose(fd);
|
236
|
3281 }
|
|
3282
|
|
3283 /*
|
300
|
3284 * Dump a word tree at node "node".
|
|
3285 *
|
|
3286 * This first writes the list of possible bytes (siblings). Then for each
|
|
3287 * byte recursively write the children.
|
|
3288 *
|
|
3289 * NOTE: The code here must match the code in read_tree(), since assumptions
|
|
3290 * are made about the indexes (so that we don't have to write them in the
|
|
3291 * file).
|
236
|
3292 *
|
300
|
3293 * Returns the number of nodes used.
|
236
|
3294 */
|
300
|
3295 static int
|
|
3296 put_tree(fd, node, index, regionmask)
|
|
3297 FILE *fd; /* NULL when only counting */
|
|
3298 wordnode_T *node;
|
|
3299 int index;
|
|
3300 int regionmask;
|
236
|
3301 {
|
300
|
3302 int newindex = index;
|
|
3303 int siblingcount = 0;
|
|
3304 wordnode_T *np;
|
236
|
3305 int flags;
|
300
|
3306
|
|
3307 /* If "node" is zero the tree is empty. */
|
|
3308 if (node == NULL)
|
|
3309 return 0;
|
|
3310
|
|
3311 /* Store the index where this node is written. */
|
|
3312 node->wn_index = index;
|
236
|
3313
|
300
|
3314 /* Count the number of siblings. */
|
|
3315 for (np = node; np != NULL; np = np->wn_sibling)
|
|
3316 ++siblingcount;
|
236
|
3317
|
300
|
3318 /* Write the sibling count. */
|
|
3319 if (fd != NULL)
|
|
3320 putc(siblingcount, fd); /* <siblingcount> */
|
236
|
3321
|
300
|
3322 /* Write each sibling byte and optionally extra info. */
|
|
3323 for (np = node; np != NULL; np = np->wn_sibling)
|
236
|
3324 {
|
300
|
3325 if (np->wn_byte == 0)
|
|
3326 {
|
|
3327 if (fd != NULL)
|
|
3328 {
|
|
3329 /* For a NUL byte (end of word) instead of the byte itself
|
|
3330 * we write the flag/region items. */
|
|
3331 flags = np->wn_flags;
|
|
3332 if (regionmask != 0 && np->wn_region != regionmask)
|
|
3333 flags |= WF_REGION;
|
|
3334 if (flags == 0)
|
|
3335 {
|
|
3336 /* word without flags or region */
|
|
3337 putc(BY_NOFLAGS, fd); /* <byte> */
|
|
3338 }
|
|
3339 else
|
|
3340 {
|
|
3341 putc(BY_FLAGS, fd); /* <byte> */
|
|
3342 putc(flags, fd); /* <flags> */
|
|
3343 if (flags & WF_REGION)
|
|
3344 putc(np->wn_region, fd); /* <regionmask> */
|
|
3345 }
|
|
3346 }
|
|
3347 }
|
|
3348 else
|
|
3349 {
|
|
3350 if (np->wn_child->wn_index != 0 && np->wn_child->wn_wnode != node)
|
|
3351 {
|
|
3352 /* The child is written elsewhere, write the reference. */
|
|
3353 if (fd != NULL)
|
|
3354 {
|
|
3355 putc(BY_INDEX, fd); /* <byte> */
|
|
3356 /* <nodeidx> */
|
|
3357 put_bytes(fd, (long_u)np->wn_child->wn_index, 3);
|
|
3358 }
|
|
3359 }
|
|
3360 else if (np->wn_child->wn_wnode == NULL)
|
|
3361 /* We will write the child below and give it an index. */
|
|
3362 np->wn_child->wn_wnode = node;
|
236
|
3363
|
300
|
3364 if (fd != NULL)
|
|
3365 if (putc(np->wn_byte, fd) == EOF) /* <byte> or <xbyte> */
|
|
3366 {
|
|
3367 EMSG(_(e_write));
|
|
3368 return 0;
|
|
3369 }
|
|
3370 }
|
236
|
3371 }
|
|
3372
|
300
|
3373 /* Space used in the array when reading: one for each sibling and one for
|
|
3374 * the count. */
|
|
3375 newindex += siblingcount + 1;
|
249
|
3376
|
300
|
3377 /* Recursively dump the children of each sibling. */
|
|
3378 for (np = node; np != NULL; np = np->wn_sibling)
|
|
3379 if (np->wn_byte != 0 && np->wn_child->wn_wnode == node)
|
|
3380 newindex = put_tree(fd, np->wn_child, newindex, regionmask);
|
249
|
3381
|
300
|
3382 return newindex;
|
236
|
3383 }
|
|
3384
|
|
3385
|
|
3386 /*
|
310
|
3387 * ":mkspell [-ascii] outfile infile ..."
|
|
3388 * ":mkspell [-ascii] addfile"
|
236
|
3389 */
|
|
3390 void
|
|
3391 ex_mkspell(eap)
|
|
3392 exarg_T *eap;
|
|
3393 {
|
|
3394 int fcount;
|
|
3395 char_u **fnames;
|
310
|
3396 char_u *arg = eap->arg;
|
|
3397 int ascii = FALSE;
|
|
3398
|
|
3399 if (STRNCMP(arg, "-ascii", 6) == 0)
|
|
3400 {
|
|
3401 ascii = TRUE;
|
|
3402 arg = skipwhite(arg + 6);
|
|
3403 }
|
|
3404
|
|
3405 /* Expand all the remaining arguments (e.g., $VIMRUNTIME). */
|
|
3406 if (get_arglist_exp(arg, &fcount, &fnames) == OK)
|
|
3407 {
|
323
|
3408 mkspell(fcount, fnames, ascii, eap->forceit, FALSE);
|
310
|
3409 FreeWild(fcount, fnames);
|
|
3410 }
|
|
3411 }
|
|
3412
|
|
3413 /*
|
|
3414 * Create a Vim spell file from one or more word lists.
|
|
3415 * "fnames[0]" is the output file name.
|
|
3416 * "fnames[fcount - 1]" is the last input file name.
|
|
3417 * Exception: when "fnames[0]" ends in ".add" it's used as the input file name
|
|
3418 * and ".spl" is appended to make the output file name.
|
|
3419 */
|
|
3420 static void
|
323
|
3421 mkspell(fcount, fnames, ascii, overwrite, added_word)
|
310
|
3422 int fcount;
|
|
3423 char_u **fnames;
|
|
3424 int ascii; /* -ascii argument given */
|
|
3425 int overwrite; /* overwrite existing output file */
|
323
|
3426 int added_word; /* invoked through "zg" */
|
310
|
3427 {
|
236
|
3428 char_u fname[MAXPATHL];
|
|
3429 char_u wfname[MAXPATHL];
|
310
|
3430 char_u **innames;
|
|
3431 int incount;
|
236
|
3432 afffile_T *(afile[8]);
|
|
3433 int i;
|
|
3434 int len;
|
|
3435 struct stat st;
|
255
|
3436 int error = FALSE;
|
300
|
3437 spellinfo_T spin;
|
|
3438
|
|
3439 vim_memset(&spin, 0, sizeof(spin));
|
323
|
3440 spin.si_verbose = !added_word;
|
310
|
3441 spin.si_ascii = ascii;
|
323
|
3442 spin.si_followup = TRUE;
|
|
3443 spin.si_rem_accents = TRUE;
|
|
3444 ga_init2(&spin.si_rep, (int)sizeof(fromto_T), 20);
|
|
3445 ga_init2(&spin.si_sal, (int)sizeof(fromto_T), 20);
|
|
3446 ga_init2(&spin.si_map, (int)sizeof(char_u), 100);
|
310
|
3447
|
|
3448 /* default: fnames[0] is output file, following are input files */
|
|
3449 innames = &fnames[1];
|
|
3450 incount = fcount - 1;
|
|
3451
|
|
3452 if (fcount >= 1)
|
240
|
3453 {
|
310
|
3454 len = STRLEN(fnames[0]);
|
|
3455 if (fcount == 1 && len > 4 && STRCMP(fnames[0] + len - 4, ".add") == 0)
|
|
3456 {
|
|
3457 /* For ":mkspell path/en.latin1.add" output file is
|
|
3458 * "path/en.latin1.add.spl". */
|
|
3459 innames = &fnames[0];
|
|
3460 incount = 1;
|
|
3461 vim_snprintf((char *)wfname, sizeof(wfname), "%s.spl", fnames[0]);
|
|
3462 }
|
|
3463 else if (len > 4 && STRCMP(fnames[0] + len - 4, ".spl") == 0)
|
|
3464 {
|
|
3465 /* Name ends in ".spl", use as the file name. */
|
323
|
3466 vim_strncpy(wfname, fnames[0], sizeof(wfname) - 1);
|
310
|
3467 }
|
|
3468 else
|
|
3469 /* Name should be language, make the file name from it. */
|
|
3470 vim_snprintf((char *)wfname, sizeof(wfname), "%s.%s.spl", fnames[0],
|
|
3471 spin.si_ascii ? (char_u *)"ascii" : spell_enc());
|
|
3472
|
|
3473 /* Check for .ascii.spl. */
|
|
3474 if (strstr((char *)gettail(wfname), ".ascii.") != NULL)
|
|
3475 spin.si_ascii = TRUE;
|
|
3476
|
|
3477 /* Check for .add.spl. */
|
|
3478 if (strstr((char *)gettail(wfname), ".add.") != NULL)
|
|
3479 spin.si_add = TRUE;
|
240
|
3480 }
|
|
3481
|
310
|
3482 if (incount <= 0)
|
236
|
3483 EMSG(_(e_invarg)); /* need at least output and input names */
|
310
|
3484 else if (incount > 8)
|
236
|
3485 EMSG(_("E754: Only up to 8 regions supported"));
|
|
3486 else
|
|
3487 {
|
|
3488 /* Check for overwriting before doing things that may take a lot of
|
|
3489 * time. */
|
310
|
3490 if (!overwrite && mch_stat((char *)wfname, &st) >= 0)
|
236
|
3491 {
|
|
3492 EMSG(_(e_exists));
|
310
|
3493 return;
|
236
|
3494 }
|
310
|
3495 if (mch_isdir(wfname))
|
236
|
3496 {
|
310
|
3497 EMSG2(_(e_isadir2), wfname);
|
|
3498 return;
|
236
|
3499 }
|
|
3500
|
|
3501 /*
|
|
3502 * Init the aff and dic pointers.
|
|
3503 * Get the region names if there are more than 2 arguments.
|
|
3504 */
|
310
|
3505 for (i = 0; i < incount; ++i)
|
236
|
3506 {
|
310
|
3507 afile[i] = NULL;
|
300
|
3508
|
316
|
3509 if (incount > 1)
|
236
|
3510 {
|
310
|
3511 len = STRLEN(innames[i]);
|
|
3512 if (STRLEN(gettail(innames[i])) < 5
|
|
3513 || innames[i][len - 3] != '_')
|
236
|
3514 {
|
310
|
3515 EMSG2(_("E755: Invalid region in %s"), innames[i]);
|
|
3516 return;
|
236
|
3517 }
|
316
|
3518 spin.si_region_name[i * 2] = TOLOWER_ASC(innames[i][len - 2]);
|
|
3519 spin.si_region_name[i * 2 + 1] =
|
|
3520 TOLOWER_ASC(innames[i][len - 1]);
|
236
|
3521 }
|
|
3522 }
|
316
|
3523 spin.si_region_count = incount;
|
236
|
3524
|
310
|
3525 if (!spin.si_add)
|
|
3526 /* Clear the char type tables, don't want to use any of the
|
|
3527 * currently used spell properties. */
|
|
3528 init_spell_chartab();
|
255
|
3529
|
300
|
3530 spin.si_foldroot = wordtree_alloc(&spin.si_blocks);
|
|
3531 spin.si_keeproot = wordtree_alloc(&spin.si_blocks);
|
|
3532 if (spin.si_foldroot == NULL || spin.si_keeproot == NULL)
|
|
3533 {
|
|
3534 error = TRUE;
|
310
|
3535 return;
|
300
|
3536 }
|
|
3537
|
236
|
3538 /*
|
|
3539 * Read all the .aff and .dic files.
|
|
3540 * Text is converted to 'encoding'.
|
300
|
3541 * Words are stored in the case-folded and keep-case trees.
|
236
|
3542 */
|
310
|
3543 for (i = 0; i < incount && !error; ++i)
|
236
|
3544 {
|
300
|
3545 spin.si_conv.vc_type = CONV_NONE;
|
310
|
3546 spin.si_region = 1 << i;
|
|
3547
|
|
3548 vim_snprintf((char *)fname, sizeof(fname), "%s.aff", innames[i]);
|
300
|
3549 if (mch_stat((char *)fname, &st) >= 0)
|
|
3550 {
|
|
3551 /* Read the .aff file. Will init "spin->si_conv" based on the
|
|
3552 * "SET" line. */
|
310
|
3553 afile[i] = spell_read_aff(fname, &spin);
|
|
3554 if (afile[i] == NULL)
|
300
|
3555 error = TRUE;
|
|
3556 else
|
|
3557 {
|
|
3558 /* Read the .dic file and store the words in the trees. */
|
|
3559 vim_snprintf((char *)fname, sizeof(fname), "%s.dic",
|
310
|
3560 innames[i]);
|
|
3561 if (spell_read_dic(fname, &spin, afile[i]) == FAIL)
|
300
|
3562 error = TRUE;
|
|
3563 }
|
|
3564 }
|
|
3565 else
|
|
3566 {
|
|
3567 /* No .aff file, try reading the file as a word list. Store
|
|
3568 * the words in the trees. */
|
310
|
3569 if (spell_read_wordfile(innames[i], &spin) == FAIL)
|
300
|
3570 error = TRUE;
|
|
3571 }
|
236
|
3572
|
310
|
3573 #ifdef FEAT_MBYTE
|
236
|
3574 /* Free any conversion stuff. */
|
300
|
3575 convert_setup(&spin.si_conv, NULL, NULL);
|
310
|
3576 #endif
|
236
|
3577 }
|
|
3578
|
300
|
3579 if (!error)
|
236
|
3580 {
|
|
3581 /*
|
300
|
3582 * Remove the dummy NUL from the start of the tree root.
|
236
|
3583 */
|
300
|
3584 spin.si_foldroot = spin.si_foldroot->wn_sibling;
|
|
3585 spin.si_keeproot = spin.si_keeproot->wn_sibling;
|
236
|
3586
|
|
3587 /*
|
300
|
3588 * Combine tails in the tree.
|
236
|
3589 */
|
323
|
3590 if (!added_word || p_verbose > 2)
|
310
|
3591 {
|
323
|
3592 if (added_word)
|
310
|
3593 verbose_enter();
|
|
3594 MSG(_("Compressing word tree..."));
|
|
3595 out_flush();
|
323
|
3596 if (added_word)
|
310
|
3597 verbose_leave();
|
|
3598 }
|
|
3599 wordtree_compress(spin.si_foldroot, &spin);
|
|
3600 wordtree_compress(spin.si_keeproot, &spin);
|
236
|
3601 }
|
|
3602
|
300
|
3603 if (!error)
|
|
3604 {
|
|
3605 /*
|
|
3606 * Write the info in the spell file.
|
|
3607 */
|
323
|
3608 if (!added_word || p_verbose > 2)
|
310
|
3609 {
|
323
|
3610 if (added_word)
|
310
|
3611 verbose_enter();
|
|
3612 smsg((char_u *)_("Writing spell file %s..."), wfname);
|
|
3613 out_flush();
|
323
|
3614 if (added_word)
|
310
|
3615 verbose_leave();
|
|
3616 }
|
|
3617
|
316
|
3618 write_vim_spell(wfname, &spin);
|
310
|
3619
|
323
|
3620 if (!added_word || p_verbose > 2)
|
310
|
3621 {
|
323
|
3622 if (added_word)
|
310
|
3623 verbose_enter();
|
|
3624 MSG(_("Done!"));
|
|
3625 smsg((char_u *)_("Estimated runtime memory use: %d bytes"),
|
302
|
3626 spin.si_memtot);
|
310
|
3627 out_flush();
|
323
|
3628 if (added_word)
|
310
|
3629 verbose_leave();
|
|
3630 }
|
|
3631
|
|
3632 /* If the file is loaded need to reload it. */
|
323
|
3633 spell_reload_one(wfname, added_word);
|
300
|
3634 }
|
|
3635
|
|
3636 /* Free the allocated memory. */
|
|
3637 free_blocks(spin.si_blocks);
|
323
|
3638 ga_clear(&spin.si_rep);
|
|
3639 ga_clear(&spin.si_sal);
|
|
3640 ga_clear(&spin.si_map);
|
300
|
3641
|
|
3642 /* Free the .aff file structures. */
|
310
|
3643 for (i = 0; i < incount; ++i)
|
|
3644 if (afile[i] != NULL)
|
|
3645 spell_free_aff(afile[i]);
|
236
|
3646 }
|
310
|
3647 }
|
|
3648
|
|
3649
|
|
3650 /*
|
|
3651 * ":spellgood {word}"
|
|
3652 * ":spellwrong {word}"
|
|
3653 */
|
|
3654 void
|
|
3655 ex_spell(eap)
|
|
3656 exarg_T *eap;
|
|
3657 {
|
|
3658 spell_add_word(eap->arg, STRLEN(eap->arg), eap->cmdidx == CMD_spellwrong);
|
236
|
3659 }
|
|
3660
|
310
|
3661 /*
|
|
3662 * Add "word[len]" to 'spellfile' as a good or bad word.
|
|
3663 */
|
|
3664 void
|
|
3665 spell_add_word(word, len, bad)
|
|
3666 char_u *word;
|
|
3667 int len;
|
|
3668 int bad;
|
|
3669 {
|
|
3670 FILE *fd;
|
|
3671 buf_T *buf;
|
|
3672
|
|
3673 if (*curbuf->b_p_spf == NUL)
|
|
3674 init_spellfile();
|
|
3675 if (*curbuf->b_p_spf == NUL)
|
323
|
3676 EMSG(_("E764: 'spellfile' is not set"));
|
310
|
3677 else
|
|
3678 {
|
|
3679 /* Check that the user isn't editing the .add file somewhere. */
|
|
3680 buf = buflist_findname_exp(curbuf->b_p_spf);
|
|
3681 if (buf != NULL && buf->b_ml.ml_mfp == NULL)
|
|
3682 buf = NULL;
|
|
3683 if (buf != NULL && bufIsChanged(buf))
|
|
3684 EMSG(_(e_bufloaded));
|
|
3685 else
|
|
3686 {
|
|
3687 fd = mch_fopen((char *)curbuf->b_p_spf, "a");
|
|
3688 if (fd == NULL)
|
|
3689 EMSG2(_(e_notopen), curbuf->b_p_spf);
|
|
3690 else
|
|
3691 {
|
|
3692 if (bad)
|
|
3693 fprintf(fd, "/!%.*s\n", len, word);
|
|
3694 else
|
|
3695 fprintf(fd, "%.*s\n", len, word);
|
|
3696 fclose(fd);
|
|
3697
|
|
3698 /* Update the .add.spl file. */
|
323
|
3699 mkspell(1, &curbuf->b_p_spf, FALSE, TRUE, TRUE);
|
310
|
3700
|
|
3701 /* If the .add file is edited somewhere, reload it. */
|
|
3702 if (buf != NULL)
|
|
3703 buf_reload(buf);
|
323
|
3704
|
|
3705 redraw_all_later(NOT_VALID);
|
310
|
3706 }
|
|
3707 }
|
|
3708 }
|
|
3709 }
|
|
3710
|
|
3711 /*
|
|
3712 * Initialize 'spellfile' for the current buffer.
|
|
3713 */
|
|
3714 static void
|
|
3715 init_spellfile()
|
|
3716 {
|
|
3717 char_u buf[MAXPATHL];
|
|
3718 int l;
|
|
3719 slang_T *sl;
|
|
3720 char_u *rtp;
|
|
3721
|
|
3722 if (*curbuf->b_p_spl != NUL && curbuf->b_langp.ga_len > 0)
|
|
3723 {
|
|
3724 /* Loop over all entries in 'runtimepath'. */
|
|
3725 rtp = p_rtp;
|
|
3726 while (*rtp != NUL)
|
|
3727 {
|
|
3728 /* Copy the path from 'runtimepath' to buf[]. */
|
|
3729 copy_option_part(&rtp, buf, MAXPATHL, ",");
|
|
3730 if (filewritable(buf) == 2)
|
|
3731 {
|
316
|
3732 /* Use the first language name from 'spelllang' and the
|
|
3733 * encoding used in the first loaded .spl file. */
|
310
|
3734 sl = LANGP_ENTRY(curbuf->b_langp, 0)->lp_slang;
|
|
3735 l = STRLEN(buf);
|
|
3736 vim_snprintf((char *)buf + l, MAXPATHL - l,
|
316
|
3737 "/spell/%.*s.%s.add",
|
|
3738 2, curbuf->b_p_spl,
|
310
|
3739 strstr((char *)gettail(sl->sl_fname), ".ascii.") != NULL
|
|
3740 ? (char_u *)"ascii" : spell_enc());
|
|
3741 set_option_value((char_u *)"spellfile", 0L, buf, OPT_LOCAL);
|
|
3742 break;
|
|
3743 }
|
|
3744 }
|
|
3745 }
|
|
3746 }
|
236
|
3747
|
300
|
3748
|
307
|
3749 /*
|
|
3750 * Init the chartab used for spelling for ASCII.
|
|
3751 * EBCDIC is not supported!
|
|
3752 */
|
|
3753 static void
|
|
3754 clear_spell_chartab(sp)
|
|
3755 spelltab_T *sp;
|
|
3756 {
|
|
3757 int i;
|
|
3758
|
|
3759 /* Init everything to FALSE. */
|
|
3760 vim_memset(sp->st_isw, FALSE, sizeof(sp->st_isw));
|
|
3761 vim_memset(sp->st_isu, FALSE, sizeof(sp->st_isu));
|
|
3762 for (i = 0; i < 256; ++i)
|
|
3763 sp->st_fold[i] = i;
|
|
3764
|
|
3765 /* We include digits. A word shouldn't start with a digit, but handling
|
|
3766 * that is done separately. */
|
|
3767 for (i = '0'; i <= '9'; ++i)
|
|
3768 sp->st_isw[i] = TRUE;
|
|
3769 for (i = 'A'; i <= 'Z'; ++i)
|
|
3770 {
|
|
3771 sp->st_isw[i] = TRUE;
|
|
3772 sp->st_isu[i] = TRUE;
|
|
3773 sp->st_fold[i] = i + 0x20;
|
|
3774 }
|
|
3775 for (i = 'a'; i <= 'z'; ++i)
|
|
3776 sp->st_isw[i] = TRUE;
|
|
3777 }
|
|
3778
|
|
3779 /*
|
|
3780 * Init the chartab used for spelling. Only depends on 'encoding'.
|
|
3781 * Called once while starting up and when 'encoding' changes.
|
|
3782 * The default is to use isalpha(), but the spell file should define the word
|
|
3783 * characters to make it possible that 'encoding' differs from the current
|
|
3784 * locale.
|
|
3785 */
|
|
3786 void
|
|
3787 init_spell_chartab()
|
|
3788 {
|
|
3789 int i;
|
|
3790
|
|
3791 did_set_spelltab = FALSE;
|
|
3792 clear_spell_chartab(&spelltab);
|
|
3793
|
|
3794 #ifdef FEAT_MBYTE
|
|
3795 if (enc_dbcs)
|
|
3796 {
|
|
3797 /* DBCS: assume double-wide characters are word characters. */
|
|
3798 for (i = 128; i <= 255; ++i)
|
|
3799 if (MB_BYTE2LEN(i) == 2)
|
|
3800 spelltab.st_isw[i] = TRUE;
|
|
3801 }
|
|
3802 else
|
|
3803 #endif
|
|
3804 {
|
|
3805 /* Rough guess: use isalpha() and isupper() for characters above 128. */
|
|
3806 for (i = 128; i < 256; ++i)
|
|
3807 {
|
|
3808 spelltab.st_isw[i] = MB_ISUPPER(i) || MB_ISLOWER(i);
|
|
3809 if (MB_ISUPPER(i))
|
|
3810 {
|
|
3811 spelltab.st_isu[i] = TRUE;
|
|
3812 spelltab.st_fold[i] = MB_TOLOWER(i);
|
|
3813 }
|
|
3814 }
|
|
3815 }
|
|
3816 }
|
|
3817
|
|
3818 static char *e_affform = N_("E761: Format error in affix file FOL, LOW or UPP");
|
|
3819 static char *e_affrange = N_("E762: Character in FOL, LOW or UPP is out of range");
|
|
3820
|
|
3821 /*
|
|
3822 * Set the spell character tables from strings in the affix file.
|
|
3823 */
|
|
3824 static int
|
|
3825 set_spell_chartab(fol, low, upp)
|
|
3826 char_u *fol;
|
|
3827 char_u *low;
|
|
3828 char_u *upp;
|
|
3829 {
|
|
3830 /* We build the new tables here first, so that we can compare with the
|
|
3831 * previous one. */
|
|
3832 spelltab_T new_st;
|
|
3833 char_u *pf = fol, *pl = low, *pu = upp;
|
|
3834 int f, l, u;
|
|
3835
|
|
3836 clear_spell_chartab(&new_st);
|
|
3837
|
|
3838 while (*pf != NUL)
|
|
3839 {
|
|
3840 if (*pl == NUL || *pu == NUL)
|
|
3841 {
|
|
3842 EMSG(_(e_affform));
|
|
3843 return FAIL;
|
|
3844 }
|
|
3845 #ifdef FEAT_MBYTE
|
|
3846 f = mb_ptr2char_adv(&pf);
|
|
3847 l = mb_ptr2char_adv(&pl);
|
|
3848 u = mb_ptr2char_adv(&pu);
|
|
3849 #else
|
|
3850 f = *pf++;
|
|
3851 l = *pl++;
|
|
3852 u = *pu++;
|
|
3853 #endif
|
|
3854 /* Every character that appears is a word character. */
|
|
3855 if (f < 256)
|
|
3856 new_st.st_isw[f] = TRUE;
|
|
3857 if (l < 256)
|
|
3858 new_st.st_isw[l] = TRUE;
|
|
3859 if (u < 256)
|
|
3860 new_st.st_isw[u] = TRUE;
|
|
3861
|
|
3862 /* if "LOW" and "FOL" are not the same the "LOW" char needs
|
|
3863 * case-folding */
|
|
3864 if (l < 256 && l != f)
|
|
3865 {
|
|
3866 if (f >= 256)
|
|
3867 {
|
|
3868 EMSG(_(e_affrange));
|
|
3869 return FAIL;
|
|
3870 }
|
|
3871 new_st.st_fold[l] = f;
|
|
3872 }
|
|
3873
|
|
3874 /* if "UPP" and "FOL" are not the same the "UPP" char needs
|
|
3875 * case-folding and it's upper case. */
|
|
3876 if (u < 256 && u != f)
|
|
3877 {
|
|
3878 if (f >= 256)
|
|
3879 {
|
|
3880 EMSG(_(e_affrange));
|
|
3881 return FAIL;
|
|
3882 }
|
|
3883 new_st.st_fold[u] = f;
|
|
3884 new_st.st_isu[u] = TRUE;
|
|
3885 }
|
|
3886 }
|
|
3887
|
|
3888 if (*pl != NUL || *pu != NUL)
|
|
3889 {
|
|
3890 EMSG(_(e_affform));
|
|
3891 return FAIL;
|
|
3892 }
|
|
3893
|
|
3894 return set_spell_finish(&new_st);
|
|
3895 }
|
|
3896
|
|
3897 /*
|
|
3898 * Set the spell character tables from strings in the .spl file.
|
|
3899 */
|
|
3900 static int
|
|
3901 set_spell_charflags(flags, cnt, upp)
|
|
3902 char_u *flags;
|
|
3903 int cnt;
|
|
3904 char_u *upp;
|
|
3905 {
|
|
3906 /* We build the new tables here first, so that we can compare with the
|
|
3907 * previous one. */
|
|
3908 spelltab_T new_st;
|
|
3909 int i;
|
|
3910 char_u *p = upp;
|
|
3911
|
|
3912 clear_spell_chartab(&new_st);
|
|
3913
|
|
3914 for (i = 0; i < cnt; ++i)
|
|
3915 {
|
|
3916 new_st.st_isw[i + 128] = (flags[i] & SPELL_ISWORD) != 0;
|
|
3917 new_st.st_isu[i + 128] = (flags[i] & SPELL_ISUPPER) != 0;
|
|
3918
|
|
3919 if (*p == NUL)
|
|
3920 return FAIL;
|
|
3921 #ifdef FEAT_MBYTE
|
|
3922 new_st.st_fold[i + 128] = mb_ptr2char_adv(&p);
|
|
3923 #else
|
|
3924 new_st.st_fold[i + 128] = *p++;
|
|
3925 #endif
|
|
3926 }
|
|
3927
|
|
3928 return set_spell_finish(&new_st);
|
|
3929 }
|
|
3930
|
|
3931 static int
|
|
3932 set_spell_finish(new_st)
|
|
3933 spelltab_T *new_st;
|
|
3934 {
|
|
3935 int i;
|
|
3936
|
|
3937 if (did_set_spelltab)
|
|
3938 {
|
|
3939 /* check that it's the same table */
|
|
3940 for (i = 0; i < 256; ++i)
|
|
3941 {
|
|
3942 if (spelltab.st_isw[i] != new_st->st_isw[i]
|
|
3943 || spelltab.st_isu[i] != new_st->st_isu[i]
|
|
3944 || spelltab.st_fold[i] != new_st->st_fold[i])
|
|
3945 {
|
|
3946 EMSG(_("E763: Word characters differ between spell files"));
|
|
3947 return FAIL;
|
|
3948 }
|
|
3949 }
|
|
3950 }
|
|
3951 else
|
|
3952 {
|
|
3953 /* copy the new spelltab into the one being used */
|
|
3954 spelltab = *new_st;
|
|
3955 did_set_spelltab = TRUE;
|
|
3956 }
|
|
3957
|
|
3958 return OK;
|
|
3959 }
|
|
3960
|
|
3961 /*
|
|
3962 * Write the current tables into the .spl file.
|
|
3963 * This makes sure the same characters are recognized as word characters when
|
|
3964 * generating an when using a spell file.
|
|
3965 */
|
|
3966 static void
|
|
3967 write_spell_chartab(fd)
|
|
3968 FILE *fd;
|
|
3969 {
|
|
3970 char_u charbuf[256 * 4];
|
|
3971 int len = 0;
|
|
3972 int flags;
|
|
3973 int i;
|
|
3974
|
|
3975 fputc(128, fd); /* <charflagslen> */
|
|
3976 for (i = 128; i < 256; ++i)
|
|
3977 {
|
|
3978 flags = 0;
|
|
3979 if (spelltab.st_isw[i])
|
|
3980 flags |= SPELL_ISWORD;
|
|
3981 if (spelltab.st_isu[i])
|
|
3982 flags |= SPELL_ISUPPER;
|
|
3983 fputc(flags, fd); /* <charflags> */
|
|
3984
|
310
|
3985 #ifdef FEAT_MBYTE
|
|
3986 if (has_mbyte)
|
|
3987 len += mb_char2bytes(spelltab.st_fold[i], charbuf + len);
|
|
3988 else
|
|
3989 #endif
|
|
3990 charbuf[len++] = spelltab.st_fold[i];
|
307
|
3991 }
|
|
3992
|
|
3993 put_bytes(fd, (long_u)len, 2); /* <fcharlen> */
|
|
3994 fwrite(charbuf, (size_t)len, (size_t)1, fd); /* <fchars> */
|
|
3995 }
|
|
3996
|
|
3997 /*
|
|
3998 * Return TRUE if "c" is an upper-case character for spelling.
|
|
3999 */
|
|
4000 static int
|
|
4001 spell_isupper(c)
|
|
4002 int c;
|
|
4003 {
|
|
4004 # ifdef FEAT_MBYTE
|
|
4005 if (enc_utf8)
|
|
4006 {
|
|
4007 /* For Unicode we can call utf_isupper(), but don't do that for ASCII,
|
|
4008 * because we don't want to use 'casemap' here. */
|
|
4009 if (c >= 128)
|
|
4010 return utf_isupper(c);
|
|
4011 }
|
|
4012 else if (has_mbyte && c > 256)
|
|
4013 {
|
|
4014 /* For characters above 255 we don't have something specfied.
|
|
4015 * Fall back to locale-dependent iswupper(). If not available
|
|
4016 * simply return FALSE. */
|
|
4017 # ifdef HAVE_ISWUPPER
|
|
4018 return iswupper(c);
|
|
4019 # else
|
|
4020 return FALSE;
|
|
4021 # endif
|
|
4022 }
|
|
4023 # endif
|
|
4024 return spelltab.st_isu[c];
|
|
4025 }
|
|
4026
|
|
4027 /*
|
|
4028 * Case-fold "p[len]" into "buf[buflen]". Used for spell checking.
|
|
4029 * When using a multi-byte 'encoding' the length may change!
|
|
4030 * Returns FAIL when something wrong.
|
|
4031 */
|
|
4032 static int
|
|
4033 spell_casefold(p, len, buf, buflen)
|
|
4034 char_u *p;
|
|
4035 int len;
|
|
4036 char_u *buf;
|
|
4037 int buflen;
|
|
4038 {
|
|
4039 int i;
|
|
4040
|
|
4041 if (len >= buflen)
|
|
4042 {
|
|
4043 buf[0] = NUL;
|
|
4044 return FAIL; /* result will not fit */
|
|
4045 }
|
|
4046
|
|
4047 #ifdef FEAT_MBYTE
|
|
4048 if (has_mbyte)
|
|
4049 {
|
|
4050 int c;
|
|
4051 int outi = 0;
|
|
4052
|
|
4053 /* Fold one character at a time. */
|
|
4054 for (i = 0; i < len; i += mb_ptr2len_check(p + i))
|
|
4055 {
|
|
4056 c = mb_ptr2char(p + i);
|
|
4057 if (enc_utf8)
|
|
4058 /* For Unicode case folding is always the same, no need to use
|
|
4059 * the table from the spell file. */
|
|
4060 c = utf_fold(c);
|
|
4061 else if (c < 256)
|
|
4062 /* Use the table from the spell file. */
|
|
4063 c = spelltab.st_fold[c];
|
|
4064 # ifdef HAVE_TOWLOWER
|
|
4065 else
|
|
4066 /* We don't know what to do, fall back to towlower(), it
|
|
4067 * depends on the current locale. */
|
|
4068 c = towlower(c);
|
|
4069 # endif
|
|
4070 if (outi + MB_MAXBYTES > buflen)
|
|
4071 {
|
|
4072 buf[outi] = NUL;
|
|
4073 return FAIL;
|
|
4074 }
|
|
4075 outi += mb_char2bytes(c, buf + outi);
|
|
4076 }
|
|
4077 buf[outi] = NUL;
|
|
4078 }
|
|
4079 else
|
|
4080 #endif
|
|
4081 {
|
|
4082 /* Be quick for non-multibyte encodings. */
|
|
4083 for (i = 0; i < len; ++i)
|
|
4084 buf[i] = spelltab.st_fold[p[i]];
|
|
4085 buf[i] = NUL;
|
|
4086 }
|
|
4087
|
|
4088 return OK;
|
|
4089 }
|
|
4090
|
323
|
4091 /*
|
|
4092 * "z?": Find badly spelled word under or after the cursor.
|
|
4093 * Give suggestions for the properly spelled word.
|
|
4094 * This is based on the mechanisms of Aspell, but completely reimplemented.
|
|
4095 */
|
|
4096 void
|
|
4097 spell_suggest()
|
|
4098 {
|
|
4099 char_u *line;
|
|
4100 pos_T prev_cursor = curwin->w_cursor;
|
|
4101 int attr;
|
|
4102 char_u wcopy[MAXWLEN + 2];
|
|
4103 char_u *p;
|
|
4104 int i;
|
|
4105 int c;
|
|
4106 suginfo_T sug;
|
|
4107 suggest_T *stp;
|
|
4108
|
|
4109 /*
|
|
4110 * Find the start of the badly spelled word.
|
|
4111 */
|
|
4112 if (spell_move_to(FORWARD, TRUE, TRUE) == FAIL)
|
|
4113 {
|
|
4114 beep_flush();
|
|
4115 return;
|
|
4116 }
|
|
4117
|
|
4118 /*
|
|
4119 * Set the info in "sug".
|
|
4120 */
|
|
4121 vim_memset(&sug, 0, sizeof(sug));
|
|
4122 ga_init2(&sug.su_ga, (int)sizeof(suggest_T), 10);
|
|
4123 hash_init(&sug.su_banned);
|
|
4124 line = ml_get_curline();
|
|
4125 sug.su_badptr = line + curwin->w_cursor.col;
|
|
4126 sug.su_badlen = spell_check(curwin, sug.su_badptr, &attr);
|
|
4127 if (sug.su_badlen >= MAXWLEN)
|
|
4128 sug.su_badlen = MAXWLEN - 1; /* just in case */
|
|
4129 vim_strncpy(sug.su_badword, sug.su_badptr, sug.su_badlen);
|
|
4130 (void)spell_casefold(sug.su_badptr, sug.su_badlen,
|
|
4131 sug.su_fbadword, MAXWLEN);
|
|
4132
|
|
4133 /* Ban the bad word itself. It may appear in another region. */
|
|
4134 add_banned(&sug, sug.su_badword);
|
|
4135
|
|
4136 /*
|
|
4137 * 1. Try inserting/deleting/swapping/changing a letter, use REP entries
|
|
4138 * from the .aff file and inserting a space (split the word).
|
|
4139 */
|
|
4140 /* Set a maximum score to limit the combination of operations that is
|
|
4141 * tried. */
|
|
4142 sug.su_maxscore = SCORE_MAXINIT;
|
|
4143 spell_try_change(&sug);
|
|
4144 cleanup_suggestions(&sug);
|
|
4145
|
|
4146 /*
|
|
4147 * 2. Try finding sound-a-like words.
|
|
4148 */
|
|
4149 /* Allow a higher score if we don't have many suggestions yet. */
|
|
4150 if (sug.su_maxscore == SCORE_MAXINIT)
|
|
4151 sug.su_maxscore = SCORE_MAXMAX;
|
|
4152 spell_try_soundalike(&sug);
|
|
4153
|
|
4154 /* When CTRL-C was hit while searching do show the results. */
|
|
4155 if (got_int)
|
|
4156 {
|
|
4157 (void)vgetc();
|
|
4158 got_int = FALSE;
|
|
4159 }
|
|
4160
|
|
4161 if (sug.su_ga.ga_len == 0)
|
|
4162 MSG(_("Sorry, no suggestions"));
|
|
4163 else
|
|
4164 {
|
|
4165 /* Cleanup, sort the suggestions and truncate at SUG_PROMPT_COUNT. */
|
|
4166 cleanup_suggestions(&sug);
|
|
4167
|
|
4168 /* List the suggestions. */
|
|
4169 msg_start();
|
|
4170 vim_snprintf((char *)IObuff, IOSIZE, _("Change \"%.*s\" to:"),
|
|
4171 sug.su_badlen, sug.su_badptr);
|
|
4172 msg_puts(IObuff);
|
|
4173 msg_clr_eos();
|
|
4174 msg_putchar('\n');
|
|
4175 msg_scroll = TRUE;
|
|
4176 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
4177 {
|
|
4178 stp = &SUG(&sug, i);
|
|
4179
|
|
4180 /* The suggested word may replace only part of the bad word, add
|
|
4181 * the not replaced part. */
|
|
4182 STRCPY(wcopy, stp->st_word);
|
|
4183 if (sug.su_badlen > stp->st_orglen)
|
|
4184 vim_strncpy(wcopy + STRLEN(wcopy),
|
|
4185 sug.su_badptr + stp->st_orglen,
|
|
4186 sug.su_badlen - stp->st_orglen);
|
|
4187 /* TODO: remove score */
|
|
4188 vim_snprintf((char *)IObuff, IOSIZE, _("%2d \"%s\" (%d)"),
|
|
4189 i + 1, wcopy, stp->st_score);
|
|
4190 msg_puts(IObuff);
|
|
4191 lines_left = 3; /* avoid more prompt */
|
|
4192 msg_putchar('\n');
|
|
4193 }
|
|
4194
|
|
4195 /* Ask for choice. */
|
|
4196 i = prompt_for_number();
|
|
4197 if (i > 0 && i <= sug.su_ga.ga_len && u_save_cursor())
|
|
4198 {
|
|
4199 /* Replace the word. */
|
|
4200 stp = &SUG(&sug, i - 1);
|
|
4201 p = alloc(STRLEN(line) - stp->st_orglen + STRLEN(stp->st_word) + 1);
|
|
4202 if (p != NULL)
|
|
4203 {
|
|
4204 c = sug.su_badptr - line;
|
|
4205 mch_memmove(p, line, c);
|
|
4206 STRCPY(p + c, stp->st_word);
|
|
4207 STRCAT(p, sug.su_badptr + stp->st_orglen);
|
|
4208 ml_replace(curwin->w_cursor.lnum, p, FALSE);
|
|
4209 curwin->w_cursor.col = c;
|
|
4210 changed_bytes(curwin->w_cursor.lnum, c);
|
|
4211 }
|
|
4212 }
|
|
4213 else
|
|
4214 curwin->w_cursor = prev_cursor;
|
|
4215 }
|
|
4216
|
|
4217 /* Free the suggestions. */
|
|
4218 for (i = 0; i < sug.su_ga.ga_len; ++i)
|
|
4219 vim_free(SUG(&sug, i).st_word);
|
|
4220 ga_clear(&sug.su_ga);
|
|
4221
|
|
4222 /* Free the banned words. */
|
|
4223 free_banned(&sug);
|
|
4224 }
|
|
4225
|
|
4226 /*
|
|
4227 * Make a copy of "word[len]", with the first letter upper or lower cased,
|
|
4228 * to "wcopy[MAXWLEN]".
|
|
4229 */
|
|
4230 static void
|
|
4231 onecap_copy(word, len, wcopy, upper)
|
|
4232 char_u *word;
|
|
4233 int len;
|
|
4234 char_u *wcopy;
|
|
4235 int upper; /* TRUE: first letter made upper case */
|
|
4236 {
|
|
4237 char_u *p;
|
|
4238 int c;
|
|
4239 int l;
|
|
4240
|
|
4241 p = word;
|
|
4242 #ifdef FEAT_MBYTE
|
|
4243 if (has_mbyte)
|
|
4244 c = mb_ptr2char_adv(&p);
|
|
4245 else
|
|
4246 #endif
|
|
4247 c = *p++;
|
|
4248 if (upper)
|
|
4249 c = MB_TOUPPER(c);
|
|
4250 else
|
|
4251 c = MB_TOLOWER(c);
|
|
4252 #ifdef FEAT_MBYTE
|
|
4253 if (has_mbyte)
|
|
4254 l = mb_char2bytes(c, wcopy);
|
|
4255 else
|
|
4256 #endif
|
|
4257 {
|
|
4258 l = 1;
|
|
4259 wcopy[0] = c;
|
|
4260 }
|
|
4261 vim_strncpy(wcopy + l, p, len - (p - word));
|
|
4262 }
|
|
4263
|
|
4264 /*
|
|
4265 * Make a copy of "word[len]" with all the letters upper cased into
|
|
4266 * "wcopy[MAXWLEN]".
|
|
4267 */
|
|
4268 static void
|
|
4269 allcap_copy(word, wcopy)
|
|
4270 char_u *word;
|
|
4271 char_u *wcopy;
|
|
4272 {
|
|
4273 char_u *s;
|
|
4274 char_u *d;
|
|
4275 int c;
|
|
4276
|
|
4277 d = wcopy;
|
|
4278 for (s = word; *s != NUL; )
|
|
4279 {
|
|
4280 #ifdef FEAT_MBYTE
|
|
4281 if (has_mbyte)
|
|
4282 c = mb_ptr2char_adv(&s);
|
|
4283 else
|
|
4284 #endif
|
|
4285 c = *s++;
|
|
4286
|
|
4287 c = MB_TOUPPER(c); /* TODO: use spell toupper */
|
|
4288
|
|
4289 #ifdef FEAT_MBYTE
|
|
4290 if (has_mbyte)
|
|
4291 {
|
|
4292 if (d - wcopy >= MAXWLEN - MB_MAXBYTES)
|
|
4293 break;
|
|
4294 d += mb_char2bytes(c, d);
|
|
4295 }
|
|
4296 else
|
|
4297 #endif
|
|
4298 {
|
|
4299 if (d - wcopy >= MAXWLEN - 1)
|
|
4300 break;
|
|
4301 *d++ = c;
|
|
4302 }
|
|
4303 }
|
|
4304 *d = NUL;
|
|
4305 }
|
|
4306
|
|
4307 /*
|
|
4308 * Try finding suggestions by adding/removing/swapping letters.
|
|
4309 */
|
|
4310 static void
|
|
4311 spell_try_change(su)
|
|
4312 suginfo_T *su;
|
|
4313 {
|
|
4314 char_u fword[MAXWLEN]; /* copy of the bad word, case-folded */
|
|
4315 char_u tword[MAXWLEN]; /* good word collected so far */
|
|
4316 trystate_T stack[MAXWLEN];
|
|
4317 char_u preword[MAXWLEN * 3]; /* word found with proper case (appended
|
|
4318 * to for word split) */
|
|
4319 char_u prewordlen = 0; /* length of word in "preword" */
|
|
4320 int splitoff = 0; /* index in tword after last split */
|
|
4321 trystate_T *sp;
|
|
4322 int newscore;
|
|
4323 langp_T *lp;
|
|
4324 char_u *byts;
|
|
4325 int *idxs;
|
|
4326 int depth;
|
|
4327 int c;
|
|
4328 int n;
|
|
4329 int flags;
|
|
4330 int badflags;
|
|
4331 garray_T *gap;
|
|
4332 int arridx;
|
|
4333 int len;
|
|
4334 char_u *p;
|
|
4335 fromto_T *ftp;
|
|
4336 int fl, tl;
|
|
4337
|
|
4338 /* get caps flags for bad word */
|
|
4339 badflags = captype(su->su_badptr, su->su_badptr + su->su_badlen);
|
|
4340
|
|
4341 /* We make a copy of the case-folded bad word, so that we can modify it
|
|
4342 * to find matches (esp. REP items). */
|
|
4343 STRCPY(fword, su->su_fbadword);
|
|
4344
|
|
4345 /*
|
|
4346 * At each node in the tree these states are tried:
|
|
4347 */
|
|
4348 #define STATE_START 0 /* At start of node, check if word may end or
|
|
4349 * split word. */
|
|
4350 #define STATE_SPLITUNDO 1 /* Undo word split. */
|
|
4351 #define STATE_ENDNUL 2 /* Past NUL bytes at start of the node. */
|
|
4352 #define STATE_PLAIN 3 /* Use each byte of the node. */
|
|
4353 #define STATE_DEL 4 /* Delete a byte from the bad word. */
|
|
4354 #define STATE_INS 5 /* Insert a byte in the bad word. */
|
|
4355 #define STATE_SWAP 6 /* Swap two bytes. */
|
|
4356 #define STATE_SWAP3A 7 /* Swap two bytes over three. */
|
|
4357 #define STATE_ROT3L 8 /* Rotate three bytes left */
|
|
4358 #define STATE_ROT3R 9 /* Rotate three bytes right */
|
|
4359 #define STATE_ROT_UNDO 10 /* undo rotating */
|
|
4360 #define STATE_REP_INI 11 /* Prepare for using REP items. */
|
|
4361 #define STATE_REP 12 /* Use matching REP items from the .aff file. */
|
|
4362 #define STATE_REP_UNDO 13 /* Undo a REP item replacement. */
|
|
4363 #define STATE_FINAL 99 /* End of this node. */
|
|
4364
|
|
4365
|
|
4366 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
4367 lp->lp_slang != NULL; ++lp)
|
|
4368 {
|
|
4369 #ifdef SOUNDFOLD_SCORE
|
|
4370 su->su_slang = lp->lp_slang;
|
|
4371 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
4372 /* soundfold the bad word */
|
|
4373 spell_soundfold(lp->lp_slang, su->su_fbadword, su->su_salword);
|
|
4374 #endif
|
|
4375
|
|
4376 /*
|
|
4377 * Go through the whole case-fold tree, try changes at each node.
|
|
4378 * "tword[]" contains the word collected from nodes in the tree.
|
|
4379 * "fword[]" the word we are trying to match with (initially the bad
|
|
4380 * word).
|
|
4381 */
|
|
4382 byts = lp->lp_slang->sl_fbyts;
|
|
4383 idxs = lp->lp_slang->sl_fidxs;
|
|
4384
|
|
4385 depth = 0;
|
|
4386 stack[0].ts_state = STATE_START;
|
|
4387 stack[0].ts_score = 0;
|
|
4388 stack[0].ts_curi = 1;
|
|
4389 stack[0].ts_fidx = 0;
|
|
4390 stack[0].ts_fidxtry = 0;
|
|
4391 stack[0].ts_twordlen = 0;
|
|
4392 stack[0].ts_arridx = 0;
|
|
4393
|
|
4394 while (depth >= 0 && !got_int)
|
|
4395 {
|
|
4396 sp = &stack[depth];
|
|
4397 switch (sp->ts_state)
|
|
4398 {
|
|
4399 case STATE_START:
|
|
4400 /*
|
|
4401 * Start of node: Deal with NUL bytes, which means
|
|
4402 * tword[] may end here.
|
|
4403 */
|
|
4404 arridx = sp->ts_arridx; /* current node in the tree */
|
|
4405 len = byts[arridx]; /* bytes in this node */
|
|
4406 arridx += sp->ts_curi; /* index of current byte */
|
|
4407
|
|
4408 if (sp->ts_curi > len || (c = byts[arridx]) != 0)
|
|
4409 {
|
|
4410 /* Past bytes in node and/or past NUL bytes. */
|
|
4411 sp->ts_state = STATE_ENDNUL;
|
|
4412 break;
|
|
4413 }
|
|
4414
|
|
4415 /*
|
|
4416 * End of word in tree.
|
|
4417 */
|
|
4418 ++sp->ts_curi; /* eat one NUL byte */
|
|
4419
|
|
4420 flags = idxs[arridx];
|
|
4421
|
|
4422 /*
|
|
4423 * Form the word with proper case in preword.
|
|
4424 * If there is a word from a previous split, append.
|
|
4425 */
|
|
4426 tword[sp->ts_twordlen] = NUL;
|
|
4427 if (flags & WF_KEEPCAP)
|
|
4428 /* Must find the word in the keep-case tree. */
|
|
4429 find_keepcap_word(lp->lp_slang, tword + splitoff,
|
|
4430 preword + prewordlen);
|
|
4431 else
|
|
4432 /* Include badflags: if the badword is onecap or allcap
|
|
4433 * use that for the goodword too. */
|
|
4434 make_case_word(tword + splitoff,
|
|
4435 preword + prewordlen, flags | badflags);
|
|
4436
|
|
4437 /* Don't use a banned word. It may appear again as a good
|
|
4438 * word, thus remember it. */
|
|
4439 if (flags & WF_BANNED)
|
|
4440 {
|
|
4441 add_banned(su, preword + prewordlen);
|
|
4442 break;
|
|
4443 }
|
|
4444 if (was_banned(su, preword + prewordlen))
|
|
4445 break;
|
|
4446
|
|
4447 newscore = 0;
|
|
4448 if ((flags & WF_REGION)
|
|
4449 && (((unsigned)flags >> 8) & lp->lp_region) == 0)
|
|
4450 newscore += SCORE_REGION;
|
|
4451 if (flags & WF_RARE)
|
|
4452 newscore += SCORE_RARE;
|
|
4453
|
|
4454 if (!spell_valid_case(badflags,
|
|
4455 captype(preword + prewordlen, NULL)))
|
|
4456 newscore += SCORE_ICASE;
|
|
4457
|
|
4458 if (fword[sp->ts_fidx] == 0)
|
|
4459 {
|
|
4460 /* The badword also ends: add suggestions, */
|
|
4461 add_suggestion(su, preword, sp->ts_score + newscore);
|
|
4462 }
|
|
4463 else if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
4464 {
|
|
4465 /* The word in the tree ends but the badword
|
|
4466 * continues: try inserting a space and check that a valid
|
|
4467 * words starts at fword[sp->ts_fidx]. */
|
|
4468 if (try_deeper(su, stack, depth, newscore + SCORE_SPLIT))
|
|
4469 {
|
|
4470 /* Save things to be restored at STATE_SPLITUNDO. */
|
|
4471 sp->ts_save_prewordlen = prewordlen;
|
|
4472 sp->ts_save_badflags = badflags;
|
|
4473 sp->ts_save_splitoff = splitoff;
|
|
4474
|
|
4475 /* Append a space to preword. */
|
|
4476 STRCAT(preword, " ");
|
|
4477 prewordlen = STRLEN(preword);
|
|
4478 splitoff = sp->ts_twordlen;
|
|
4479 /* TODO: when case-folding changed the number of bytes
|
|
4480 * this doesn't work... */
|
|
4481 badflags = captype(su->su_badptr + sp->ts_fidx,
|
|
4482 su->su_badptr + su->su_badlen);
|
|
4483
|
|
4484 sp->ts_state = STATE_SPLITUNDO;
|
|
4485 ++depth;
|
|
4486 /* Restart at top of the tree. */
|
|
4487 stack[depth].ts_arridx = 0;
|
|
4488 }
|
|
4489 }
|
|
4490 break;
|
|
4491
|
|
4492 case STATE_SPLITUNDO:
|
|
4493 /* Fixup the changes done for word split. */
|
|
4494 badflags = sp->ts_save_badflags;
|
|
4495 splitoff = sp->ts_save_splitoff;
|
|
4496 prewordlen = sp->ts_save_prewordlen;
|
|
4497
|
|
4498 /* Continue looking for NUL bytes. */
|
|
4499 sp->ts_state = STATE_START;
|
|
4500 break;
|
|
4501
|
|
4502 case STATE_ENDNUL:
|
|
4503 /* Past the NUL bytes in the node. */
|
|
4504 if (fword[sp->ts_fidx] == 0)
|
|
4505 {
|
|
4506 /* The badword ends, can't use the bytes in this node. */
|
|
4507 sp->ts_state = STATE_DEL;
|
|
4508 break;
|
|
4509 }
|
|
4510 sp->ts_state = STATE_PLAIN;
|
|
4511 /*FALLTHROUGH*/
|
|
4512
|
|
4513 case STATE_PLAIN:
|
|
4514 /*
|
|
4515 * Go over all possible bytes at this node, add each to
|
|
4516 * tword[] and use child node. "ts_curi" is the index.
|
|
4517 */
|
|
4518 arridx = sp->ts_arridx;
|
|
4519 if (sp->ts_curi > byts[arridx])
|
|
4520 {
|
|
4521 /* Done all bytes at this node, do next state. When still
|
|
4522 * at already changed bytes skip the other tricks. */
|
|
4523 if (sp->ts_fidx >= sp->ts_fidxtry)
|
|
4524 sp->ts_state = STATE_DEL;
|
|
4525 else
|
|
4526 sp->ts_state = STATE_FINAL;
|
|
4527 }
|
|
4528 else
|
|
4529 {
|
|
4530 arridx += sp->ts_curi++;
|
|
4531 c = byts[arridx];
|
|
4532
|
|
4533 /* Normal byte, go one level deeper. If it's not equal to
|
|
4534 * the byte in the bad word adjust the score. But don't
|
|
4535 * even try when the byte was already changed. */
|
|
4536 if (c == fword[sp->ts_fidx])
|
|
4537 newscore = 0;
|
|
4538 /* TODO: multi-byte characters */
|
|
4539 else if (lp->lp_slang->sl_map != NULL
|
|
4540 && similar_chars(lp->lp_slang,
|
|
4541 c, fword[sp->ts_fidx]))
|
|
4542 newscore = SCORE_SIMILAR;
|
|
4543 else
|
|
4544 newscore = SCORE_SUBST;
|
|
4545 if ((newscore == 0 || sp->ts_fidx >= sp->ts_fidxtry)
|
|
4546 && try_deeper(su, stack, depth, newscore))
|
|
4547 {
|
|
4548 ++depth;
|
|
4549 ++stack[depth].ts_fidx;
|
|
4550 tword[stack[depth].ts_twordlen++] = c;
|
|
4551 stack[depth].ts_arridx = idxs[arridx];
|
|
4552 }
|
|
4553 }
|
|
4554 break;
|
|
4555
|
|
4556 case STATE_DEL:
|
|
4557 /* Try skipping one byte in the bad word (delete it). */
|
|
4558 sp->ts_state = STATE_INS;
|
|
4559 sp->ts_curi = 1;
|
|
4560 if (fword[sp->ts_fidx] != NUL
|
|
4561 && try_deeper(su, stack, depth, SCORE_DEL))
|
|
4562 {
|
|
4563 ++depth;
|
|
4564 ++stack[depth].ts_fidx;
|
|
4565 break;
|
|
4566 }
|
|
4567 /*FALLTHROUGH*/
|
|
4568
|
|
4569 case STATE_INS:
|
|
4570 /* Insert one byte. Do this for each possible bytes at this
|
|
4571 * node. */
|
|
4572 n = sp->ts_arridx;
|
|
4573 if (sp->ts_curi > byts[n])
|
|
4574 {
|
|
4575 /* Done all bytes at this node, do next state. */
|
|
4576 sp->ts_state = STATE_SWAP;
|
|
4577 sp->ts_curi = 1;
|
|
4578 }
|
|
4579 else
|
|
4580 {
|
|
4581 /* Do one more byte at this node. */
|
|
4582 n += sp->ts_curi++;
|
|
4583 c = byts[n];
|
|
4584 if (c != 0 && try_deeper(su, stack, depth, SCORE_INS))
|
|
4585 {
|
|
4586 ++depth;
|
|
4587 tword[stack[depth].ts_twordlen++] = c;
|
|
4588 stack[depth].ts_arridx = idxs[n];
|
|
4589 }
|
|
4590 }
|
|
4591 break;
|
|
4592
|
|
4593 case STATE_SWAP:
|
|
4594 /* Swap two bytes: "12" -> "21". This means looking for the
|
|
4595 * following byte at the current node and the current byte at
|
|
4596 * its child node. We change "fword" here, it's changed back
|
|
4597 * afterwards. TODO: should swap characters instead of bytes.
|
|
4598 * */
|
|
4599 c = fword[sp->ts_fidx];
|
|
4600 if (c != NUL && fword[sp->ts_fidx + 1] != NUL
|
|
4601 && try_deeper(su, stack, depth, SCORE_SWAP))
|
|
4602 {
|
|
4603 sp->ts_state = STATE_SWAP3A;
|
|
4604 ++depth;
|
|
4605 fword[sp->ts_fidx] = fword[sp->ts_fidx + 1];
|
|
4606 fword[sp->ts_fidx + 1] = c;
|
|
4607 stack[depth].ts_fidxtry = sp->ts_fidx + 2;
|
|
4608 }
|
|
4609 else
|
|
4610 /* If this swap doesn't work then SWAP3 won't either. */
|
|
4611 sp->ts_state = STATE_REP_INI;
|
|
4612 break;
|
|
4613
|
|
4614 case STATE_SWAP3A:
|
|
4615 /* First undo the STATE_SWAP swap: "21" -> "12". */
|
|
4616 c = fword[sp->ts_fidx];
|
|
4617 fword[sp->ts_fidx] = fword[sp->ts_fidx + 1];
|
|
4618 fword[sp->ts_fidx + 1] = c;
|
|
4619
|
|
4620 /* Swap two bytes, skipping one: "123" -> "321". We change
|
|
4621 * "fword" here, it's changed back afterwards. TODO: should
|
|
4622 * swap characters instead of bytes. */
|
|
4623 c = fword[sp->ts_fidx];
|
|
4624 if (c != NUL && fword[sp->ts_fidx + 1] != NUL
|
|
4625 && fword[sp->ts_fidx + 2] != NUL
|
|
4626 && try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
4627 {
|
|
4628 sp->ts_state = STATE_ROT3L;
|
|
4629 ++depth;
|
|
4630 fword[sp->ts_fidx] = fword[sp->ts_fidx + 2];
|
|
4631 fword[sp->ts_fidx + 2] = c;
|
|
4632 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
4633 }
|
|
4634 else
|
|
4635 sp->ts_state = STATE_REP_INI;
|
|
4636 break;
|
|
4637
|
|
4638 case STATE_ROT3L:
|
|
4639 /* First undo STATE_SWAP3A: "321" -> "123" */
|
|
4640 c = fword[sp->ts_fidx];
|
|
4641 fword[sp->ts_fidx] = fword[sp->ts_fidx + 2];
|
|
4642 fword[sp->ts_fidx + 2] = c;
|
|
4643
|
|
4644 /* Rotate three bytes left: "123" -> "231". We change
|
|
4645 * "fword" here, it's changed back afterwards. TODO: should
|
|
4646 * swap characters instead of bytes. */
|
|
4647 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
4648 {
|
|
4649 sp->ts_state = STATE_ROT3R;
|
|
4650 ++depth;
|
|
4651 c = fword[sp->ts_fidx];
|
|
4652 fword[sp->ts_fidx] = fword[sp->ts_fidx + 1];
|
|
4653 fword[sp->ts_fidx + 1] = fword[sp->ts_fidx + 2];
|
|
4654 fword[sp->ts_fidx + 2] = c;
|
|
4655 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
4656 }
|
|
4657 else
|
|
4658 sp->ts_state = STATE_REP_INI;
|
|
4659 break;
|
|
4660
|
|
4661 case STATE_ROT3R:
|
|
4662 /* First undo STATE_ROT3L: "231" -> "123" */
|
|
4663 c = fword[sp->ts_fidx + 2];
|
|
4664 fword[sp->ts_fidx + 2] = fword[sp->ts_fidx + 1];
|
|
4665 fword[sp->ts_fidx + 1] = fword[sp->ts_fidx];
|
|
4666 fword[sp->ts_fidx] = c;
|
|
4667
|
|
4668 /* Rotate three bytes right: "123" -> "312". We change
|
|
4669 * "fword" here, it's changed back afterwards. TODO: should
|
|
4670 * swap characters instead of bytes. */
|
|
4671 if (try_deeper(su, stack, depth, SCORE_SWAP3))
|
|
4672 {
|
|
4673 sp->ts_state = STATE_ROT_UNDO;
|
|
4674 ++depth;
|
|
4675 c = fword[sp->ts_fidx + 2];
|
|
4676 fword[sp->ts_fidx + 2] = fword[sp->ts_fidx + 1];
|
|
4677 fword[sp->ts_fidx + 1] = fword[sp->ts_fidx];
|
|
4678 fword[sp->ts_fidx] = c;
|
|
4679 stack[depth].ts_fidxtry = sp->ts_fidx + 3;
|
|
4680 }
|
|
4681 else
|
|
4682 sp->ts_state = STATE_REP_INI;
|
|
4683 break;
|
|
4684
|
|
4685 case STATE_ROT_UNDO:
|
|
4686 /* Undo STATE_ROT3R: "312" -> "123" */
|
|
4687 c = fword[sp->ts_fidx];
|
|
4688 fword[sp->ts_fidx] = fword[sp->ts_fidx + 1];
|
|
4689 fword[sp->ts_fidx + 1] = fword[sp->ts_fidx + 2];
|
|
4690 fword[sp->ts_fidx + 2] = c;
|
|
4691 /*FALLTHROUGH*/
|
|
4692
|
|
4693 case STATE_REP_INI:
|
|
4694 /* Check if matching with REP items from the .aff file would
|
|
4695 * work. Quickly skip if there are no REP items or the score
|
|
4696 * is going to be too high anyway. */
|
|
4697 gap = &lp->lp_slang->sl_rep;
|
|
4698 if (gap->ga_len == 0
|
|
4699 || sp->ts_score + SCORE_REP >= su->su_maxscore)
|
|
4700 {
|
|
4701 sp->ts_state = STATE_FINAL;
|
|
4702 break;
|
|
4703 }
|
|
4704
|
|
4705 /* Use the first byte to quickly find the first entry that
|
|
4706 * matches. If the index is -1 there is none. */
|
|
4707 sp->ts_curi = lp->lp_slang->sl_rep_first[fword[sp->ts_fidx]];
|
|
4708 if (sp->ts_curi < 0)
|
|
4709 {
|
|
4710 sp->ts_state = STATE_FINAL;
|
|
4711 break;
|
|
4712 }
|
|
4713
|
|
4714 sp->ts_state = STATE_REP;
|
|
4715 /*FALLTHROUGH*/
|
|
4716
|
|
4717 case STATE_REP:
|
|
4718 /* Try matching with REP items from the .aff file. For each
|
|
4719 * match replace the charactes and check if the resulting word
|
|
4720 * is valid. */
|
|
4721 p = fword + sp->ts_fidx;
|
|
4722
|
|
4723 gap = &lp->lp_slang->sl_rep;
|
|
4724 while (sp->ts_curi < gap->ga_len)
|
|
4725 {
|
|
4726 ftp = (fromto_T *)gap->ga_data + sp->ts_curi++;
|
|
4727 if (*ftp->ft_from != *p)
|
|
4728 {
|
|
4729 /* past possible matching entries */
|
|
4730 sp->ts_curi = gap->ga_len;
|
|
4731 break;
|
|
4732 }
|
|
4733 if (STRNCMP(ftp->ft_from, p, STRLEN(ftp->ft_from)) == 0
|
|
4734 && try_deeper(su, stack, depth, SCORE_REP))
|
|
4735 {
|
|
4736 /* Need to undo this afterwards. */
|
|
4737 sp->ts_state = STATE_REP_UNDO;
|
|
4738
|
|
4739 /* Change the "from" to the "to" string. */
|
|
4740 ++depth;
|
|
4741 fl = STRLEN(ftp->ft_from);
|
|
4742 tl = STRLEN(ftp->ft_to);
|
|
4743 if (fl != tl)
|
|
4744 mch_memmove(p + tl, p + fl, STRLEN(p + fl) + 1);
|
|
4745 mch_memmove(p, ftp->ft_to, tl);
|
|
4746 stack[depth].ts_fidxtry = sp->ts_fidx + tl;
|
|
4747 break;
|
|
4748 }
|
|
4749 }
|
|
4750
|
|
4751 if (sp->ts_curi >= gap->ga_len)
|
|
4752 /* No (more) matches. */
|
|
4753 sp->ts_state = STATE_FINAL;
|
|
4754
|
|
4755 break;
|
|
4756
|
|
4757 case STATE_REP_UNDO:
|
|
4758 /* Undo a REP replacement and continue with the next one. */
|
|
4759 ftp = (fromto_T *)lp->lp_slang->sl_rep.ga_data
|
|
4760 + sp->ts_curi - 1;
|
|
4761 fl = STRLEN(ftp->ft_from);
|
|
4762 tl = STRLEN(ftp->ft_to);
|
|
4763 p = fword + sp->ts_fidx;
|
|
4764 if (fl != tl)
|
|
4765 mch_memmove(p + fl, p + tl, STRLEN(p + tl) + 1);
|
|
4766 mch_memmove(p, ftp->ft_from, fl);
|
|
4767 sp->ts_state = STATE_REP;
|
|
4768 break;
|
|
4769
|
|
4770 default:
|
|
4771 /* Did all possible states at this level, go up one level. */
|
|
4772 --depth;
|
|
4773 }
|
|
4774
|
|
4775 line_breakcheck();
|
|
4776 }
|
|
4777 }
|
|
4778 }
|
|
4779
|
|
4780 /*
|
|
4781 * Try going one level deeper in the tree.
|
|
4782 */
|
|
4783 static int
|
|
4784 try_deeper(su, stack, depth, score_add)
|
|
4785 suginfo_T *su;
|
|
4786 trystate_T *stack;
|
|
4787 int depth;
|
|
4788 int score_add;
|
|
4789 {
|
|
4790 int newscore;
|
|
4791
|
|
4792 /* Refuse to go deeper if the scrore is getting too big. */
|
|
4793 newscore = stack[depth].ts_score + score_add;
|
|
4794 if (newscore >= su->su_maxscore)
|
|
4795 return FALSE;
|
|
4796
|
|
4797 stack[depth + 1].ts_state = STATE_START;
|
|
4798 stack[depth + 1].ts_score = newscore;
|
|
4799 stack[depth + 1].ts_curi = 1; /* start just after length byte */
|
|
4800 stack[depth + 1].ts_fidx = stack[depth].ts_fidx;
|
|
4801 stack[depth + 1].ts_fidxtry = stack[depth].ts_fidxtry;
|
|
4802 stack[depth + 1].ts_twordlen = stack[depth].ts_twordlen;
|
|
4803 stack[depth + 1].ts_arridx = stack[depth].ts_arridx;
|
|
4804 return TRUE;
|
|
4805 }
|
|
4806
|
|
4807 /*
|
|
4808 * "fword" is a good word with case folded. Find the matching keep-case
|
|
4809 * words and put it in "kword".
|
|
4810 * Theoretically there could be several keep-case words that result in the
|
|
4811 * same case-folded word, but we only find one...
|
|
4812 */
|
|
4813 static void
|
|
4814 find_keepcap_word(slang, fword, kword)
|
|
4815 slang_T *slang;
|
|
4816 char_u *fword;
|
|
4817 char_u *kword;
|
|
4818 {
|
|
4819 char_u uword[MAXWLEN]; /* "fword" in upper-case */
|
|
4820 int depth;
|
|
4821 int tryidx;
|
|
4822
|
|
4823 /* The following arrays are used at each depth in the tree. */
|
|
4824 int arridx[MAXWLEN];
|
|
4825 int round[MAXWLEN];
|
|
4826 int fwordidx[MAXWLEN];
|
|
4827 int uwordidx[MAXWLEN];
|
|
4828 int kwordlen[MAXWLEN];
|
|
4829
|
|
4830 int flen, ulen;
|
|
4831 int l;
|
|
4832 int len;
|
|
4833 int c;
|
|
4834 unsigned lo, hi, m;
|
|
4835 char_u *p;
|
|
4836 char_u *byts = slang->sl_kbyts; /* array with bytes of the words */
|
|
4837 int *idxs = slang->sl_kidxs; /* array with indexes */
|
|
4838
|
|
4839 if (byts == NULL)
|
|
4840 {
|
|
4841 /* array is empty: "cannot happen" */
|
|
4842 *kword = NUL;
|
|
4843 return;
|
|
4844 }
|
|
4845
|
|
4846 /* Make an all-cap version of "fword". */
|
|
4847 allcap_copy(fword, uword);
|
|
4848
|
|
4849 /*
|
|
4850 * Each character needs to be tried both case-folded and upper-case.
|
|
4851 * All this gets very complicated if we keep in mind that changing case
|
|
4852 * may change the byte length of a multi-byte character...
|
|
4853 */
|
|
4854 depth = 0;
|
|
4855 arridx[0] = 0;
|
|
4856 round[0] = 0;
|
|
4857 fwordidx[0] = 0;
|
|
4858 uwordidx[0] = 0;
|
|
4859 kwordlen[0] = 0;
|
|
4860 while (depth >= 0)
|
|
4861 {
|
|
4862 if (fword[fwordidx[depth]] == NUL)
|
|
4863 {
|
|
4864 /* We are at the end of "fword". If the tree allows a word to end
|
|
4865 * here we have found a match. */
|
|
4866 if (byts[arridx[depth] + 1] == 0)
|
|
4867 {
|
|
4868 kword[kwordlen[depth]] = NUL;
|
|
4869 return;
|
|
4870 }
|
|
4871
|
|
4872 /* kword is getting too long, continue one level up */
|
|
4873 --depth;
|
|
4874 }
|
|
4875 else if (++round[depth] > 2)
|
|
4876 {
|
|
4877 /* tried both fold-case and upper-case character, continue one
|
|
4878 * level up */
|
|
4879 --depth;
|
|
4880 }
|
|
4881 else
|
|
4882 {
|
|
4883 /*
|
|
4884 * round[depth] == 1: Try using the folded-case character.
|
|
4885 * round[depth] == 2: Try using the upper-case character.
|
|
4886 */
|
|
4887 #ifdef FEAT_MBYTE
|
|
4888 if (has_mbyte)
|
|
4889 {
|
|
4890 flen = mb_ptr2len_check(fword + fwordidx[depth]);
|
|
4891 ulen = mb_ptr2len_check(uword + uwordidx[depth]);
|
|
4892 }
|
|
4893 else
|
|
4894 #endif
|
|
4895 ulen = flen = 1;
|
|
4896 if (round[depth] == 1)
|
|
4897 {
|
|
4898 p = fword + fwordidx[depth];
|
|
4899 l = flen;
|
|
4900 }
|
|
4901 else
|
|
4902 {
|
|
4903 p = uword + uwordidx[depth];
|
|
4904 l = ulen;
|
|
4905 }
|
|
4906
|
|
4907 for (tryidx = arridx[depth]; l > 0; --l)
|
|
4908 {
|
|
4909 /* Perform a binary search in the list of accepted bytes. */
|
|
4910 len = byts[tryidx++];
|
|
4911 c = *p++;
|
|
4912 lo = tryidx;
|
|
4913 hi = tryidx + len - 1;
|
|
4914 while (lo < hi)
|
|
4915 {
|
|
4916 m = (lo + hi) / 2;
|
|
4917 if (byts[m] > c)
|
|
4918 hi = m - 1;
|
|
4919 else if (byts[m] < c)
|
|
4920 lo = m + 1;
|
|
4921 else
|
|
4922 {
|
|
4923 lo = hi = m;
|
|
4924 break;
|
|
4925 }
|
|
4926 }
|
|
4927
|
|
4928 /* Stop if there is no matching byte. */
|
|
4929 if (hi < lo || byts[lo] != c)
|
|
4930 break;
|
|
4931
|
|
4932 /* Continue at the child (if there is one). */
|
|
4933 tryidx = idxs[lo];
|
|
4934 }
|
|
4935
|
|
4936 if (l == 0)
|
|
4937 {
|
|
4938 /*
|
|
4939 * Found the matching char. Copy it to "kword" and go a
|
|
4940 * level deeper.
|
|
4941 */
|
|
4942 if (round[depth] == 1)
|
|
4943 {
|
|
4944 STRNCPY(kword + kwordlen[depth], fword + fwordidx[depth],
|
|
4945 flen);
|
|
4946 kwordlen[depth + 1] = kwordlen[depth] + flen;
|
|
4947 }
|
|
4948 else
|
|
4949 {
|
|
4950 STRNCPY(kword + kwordlen[depth], uword + uwordidx[depth],
|
|
4951 ulen);
|
|
4952 kwordlen[depth + 1] = kwordlen[depth] + ulen;
|
|
4953 }
|
|
4954 fwordidx[depth + 1] = fwordidx[depth] + flen;
|
|
4955 uwordidx[depth + 1] = uwordidx[depth] + ulen;
|
|
4956
|
|
4957 ++depth;
|
|
4958 arridx[depth] = tryidx;
|
|
4959 round[depth] = 0;
|
|
4960 }
|
|
4961 }
|
|
4962 }
|
|
4963
|
|
4964 /* Didn't find it: "cannot happen". */
|
|
4965 *kword = NUL;
|
|
4966 }
|
|
4967
|
|
4968 /*
|
|
4969 * Find suggestions by comparing the word in a sound-a-like form.
|
|
4970 */
|
|
4971 static void
|
|
4972 spell_try_soundalike(su)
|
|
4973 suginfo_T *su;
|
|
4974 {
|
|
4975 char_u salword[MAXWLEN];
|
|
4976 char_u tword[MAXWLEN];
|
|
4977 char_u tfword[MAXWLEN];
|
|
4978 char_u tsalword[MAXWLEN];
|
|
4979 int arridx[MAXWLEN];
|
|
4980 int curi[MAXWLEN];
|
|
4981 langp_T *lp;
|
|
4982 char_u *byts;
|
|
4983 int *idxs;
|
|
4984 int depth;
|
|
4985 int c;
|
|
4986 int n;
|
|
4987 int round;
|
|
4988 int flags;
|
|
4989
|
|
4990 for (lp = LANGP_ENTRY(curwin->w_buffer->b_langp, 0);
|
|
4991 lp->lp_slang != NULL; ++lp)
|
|
4992 {
|
|
4993 if (lp->lp_slang->sl_sal.ga_len > 0)
|
|
4994 {
|
|
4995 /* soundfold the bad word */
|
|
4996 spell_soundfold(lp->lp_slang, su->su_fbadword, salword);
|
|
4997
|
|
4998 /*
|
|
4999 * Go through the whole tree, soundfold each word and compare.
|
|
5000 * round 1: use the case-folded tree.
|
|
5001 * round 2: use the keep-case tree.
|
|
5002 */
|
|
5003 for (round = 1; round <= 2; ++round)
|
|
5004 {
|
|
5005 if (round == 1)
|
|
5006 {
|
|
5007 byts = lp->lp_slang->sl_fbyts;
|
|
5008 idxs = lp->lp_slang->sl_fidxs;
|
|
5009 }
|
|
5010 else
|
|
5011 {
|
|
5012 byts = lp->lp_slang->sl_kbyts;
|
|
5013 idxs = lp->lp_slang->sl_kidxs;
|
|
5014 }
|
|
5015
|
|
5016 depth = 0;
|
|
5017 arridx[0] = 0;
|
|
5018 curi[0] = 1;
|
|
5019 while (depth >= 0 && !got_int)
|
|
5020 {
|
|
5021 if (curi[depth] > byts[arridx[depth]])
|
|
5022 /* Done all bytes at this node, go up one level. */
|
|
5023 --depth;
|
|
5024 else
|
|
5025 {
|
|
5026 /* Do one more byte at this node. */
|
|
5027 n = arridx[depth] + curi[depth];
|
|
5028 ++curi[depth];
|
|
5029 c = byts[n];
|
|
5030 if (c == 0)
|
|
5031 {
|
|
5032 /* End of word, deal with the word. */
|
|
5033 flags = idxs[n];
|
|
5034 if (round == 2 || (flags & WF_KEEPCAP) == 0)
|
|
5035 {
|
|
5036 tword[depth] = NUL;
|
|
5037 if (round == 1)
|
|
5038 spell_soundfold(lp->lp_slang,
|
|
5039 tword, tsalword);
|
|
5040 else
|
|
5041 {
|
|
5042 /* In keep-case tree need to case-fold the
|
|
5043 * word. */
|
|
5044 (void)spell_casefold(tword, depth,
|
|
5045 tfword, MAXWLEN);
|
|
5046 spell_soundfold(lp->lp_slang,
|
|
5047 tfword, tsalword);
|
|
5048 }
|
|
5049
|
|
5050 /* TODO: also compare with small changes
|
|
5051 * (insert char, swap char, etc.) */
|
|
5052 if (STRCMP(salword, tsalword) == 0)
|
|
5053 {
|
|
5054 if (round == 1 && flags != 0)
|
|
5055 {
|
|
5056 char_u cword[MAXWLEN];
|
|
5057
|
|
5058 make_case_word(tword, cword, flags);
|
|
5059 add_suggestion(su, cword, 0);
|
|
5060 }
|
|
5061 else
|
|
5062 add_suggestion(su, tword, 0);
|
|
5063 }
|
|
5064 }
|
|
5065
|
|
5066 /* Skip over other NUL bytes. */
|
|
5067 while (byts[n + 1] == 0)
|
|
5068 {
|
|
5069 ++n;
|
|
5070 ++curi[depth];
|
|
5071 }
|
|
5072 }
|
|
5073 else
|
|
5074 {
|
|
5075 /* Normal char, go one level deeper. */
|
|
5076 tword[depth++] = c;
|
|
5077 arridx[depth] = idxs[n];
|
|
5078 curi[depth] = 1;
|
|
5079 }
|
|
5080 }
|
|
5081 }
|
|
5082 line_breakcheck();
|
|
5083 }
|
|
5084 }
|
|
5085 }
|
|
5086 }
|
|
5087
|
|
5088 /*
|
|
5089 * Copy "fword" to "cword", fixing according to "flags".
|
|
5090 */
|
|
5091 static void
|
|
5092 make_case_word(fword, cword, flags)
|
|
5093 char_u *fword;
|
|
5094 char_u *cword;
|
|
5095 int flags;
|
|
5096 {
|
|
5097 if (flags & WF_ALLCAP)
|
|
5098 /* Make it all upper-case */
|
|
5099 allcap_copy(fword, cword);
|
|
5100 else if (flags & WF_ONECAP)
|
|
5101 /* Make the first letter upper-case */
|
|
5102 onecap_copy(fword, STRLEN(fword), cword, TRUE);
|
|
5103 else
|
|
5104 /* Use goodword as-is. */
|
|
5105 STRCPY(cword, fword);
|
|
5106 }
|
|
5107
|
|
5108 /*
|
|
5109 * Return TRUE if "c1" and "c2" are similar characters according to the MAP
|
|
5110 * lines in the .aff file.
|
|
5111 */
|
|
5112 static int
|
|
5113 similar_chars(slang, c1, c2)
|
|
5114 slang_T *slang;
|
|
5115 int c1;
|
|
5116 int c2;
|
|
5117 {
|
|
5118 char_u *p1;
|
|
5119 char_u *p2;
|
|
5120
|
|
5121 /* The similar characters are stored separated with slashes:
|
|
5122 * "aaa/bbb/ccc/". Search for each character and if the next slash is the
|
|
5123 * same one they are in the same MAP entry. */
|
|
5124 p1 = vim_strchr(slang->sl_map, c1);
|
|
5125 if (p1 == NULL)
|
|
5126 return FALSE;
|
|
5127 p2 = vim_strchr(slang->sl_map, c2);
|
|
5128 if (p2 == NULL)
|
|
5129 return FALSE;
|
|
5130 return vim_strchr(p1, '/') == vim_strchr(p2, '/');
|
|
5131 }
|
|
5132
|
|
5133 /*
|
|
5134 * Add a suggestion to the list of suggestions.
|
|
5135 * Do not add a duplicate suggestion or suggestions with a bad score.
|
|
5136 * When "use_score" is not zero it's used, otherwise the score is computed
|
|
5137 * with spell_edit_score().
|
|
5138 */
|
|
5139 static void
|
|
5140 add_suggestion(su, goodword, use_score)
|
|
5141 suginfo_T *su;
|
|
5142 char_u *goodword;
|
|
5143 int use_score;
|
|
5144 {
|
|
5145 suggest_T *stp;
|
|
5146 int score;
|
|
5147 int i;
|
|
5148 #ifdef SOUNDFOLD_SCORE
|
|
5149 char_u fword[MAXWLEN];
|
|
5150 char_u salword[MAXWLEN];
|
|
5151 #endif
|
|
5152
|
|
5153 /* Check that the word wasn't banned. */
|
|
5154 if (was_banned(su, goodword))
|
|
5155 return;
|
|
5156
|
|
5157 /* Compute the score and add the suggestion if it's good enough. */
|
|
5158 if (use_score != 0)
|
|
5159 score = use_score;
|
|
5160 else
|
|
5161 score = spell_edit_score(su->su_badword, goodword);
|
|
5162
|
|
5163 if (score <= su->su_maxscore)
|
|
5164 {
|
|
5165 #ifdef SOUNDFOLD_SCORE
|
|
5166 /* Add to the score when the word sounds differently.
|
|
5167 * This is slow... */
|
|
5168 if (su->su_slang->sl_sal.ga_len > 0)
|
|
5169 {
|
|
5170 (void)spell_casefold(goodword, STRLEN(goodword), fword, MAXWLEN);
|
|
5171 spell_soundfold(su->su_slang, fword, salword);
|
|
5172 score += spell_edit_score(su->su_salword, salword);
|
|
5173 }
|
|
5174 #endif
|
|
5175
|
|
5176 /* Check if the word is already there. */
|
|
5177 stp = &SUG(su, 0);
|
|
5178 for (i = su->su_ga.ga_len - 1; i >= 0; --i)
|
|
5179 if (STRCMP(stp[i].st_word, goodword) == 0)
|
|
5180 {
|
|
5181 /* Found it. Remember the lowest score. */
|
|
5182 if (stp[i].st_score > score)
|
|
5183 stp[i].st_score = score;
|
|
5184 break;
|
|
5185 }
|
|
5186
|
|
5187 if (i < 0 && ga_grow(&su->su_ga, 1) == OK)
|
|
5188 {
|
|
5189 /* Add a suggestion. */
|
|
5190 stp = &SUG(su, su->su_ga.ga_len);
|
|
5191 stp->st_word = vim_strsave(goodword);
|
|
5192 if (stp->st_word != NULL)
|
|
5193 {
|
|
5194 stp->st_score = score;
|
|
5195 stp->st_orglen = su->su_badlen;
|
|
5196 ++su->su_ga.ga_len;
|
|
5197
|
|
5198 /* If we have too many suggestions now, sort the list and keep
|
|
5199 * the best suggestions. */
|
|
5200 if (su->su_ga.ga_len > SUG_CLEANUP_COUNT)
|
|
5201 cleanup_suggestions(su);
|
|
5202 }
|
|
5203 }
|
|
5204 }
|
|
5205 }
|
|
5206
|
|
5207 /*
|
|
5208 * Add a word to be banned.
|
|
5209 */
|
|
5210 static void
|
|
5211 add_banned(su, word)
|
|
5212 suginfo_T *su;
|
|
5213 char_u *word;
|
|
5214 {
|
|
5215 char_u *s = vim_strsave(word);
|
|
5216 hash_T hash;
|
|
5217 hashitem_T *hi;
|
|
5218
|
|
5219 if (s != NULL)
|
|
5220 {
|
|
5221 hash = hash_hash(s);
|
|
5222 hi = hash_lookup(&su->su_banned, s, hash);
|
|
5223 if (HASHITEM_EMPTY(hi))
|
|
5224 hash_add_item(&su->su_banned, hi, s, hash);
|
|
5225 }
|
|
5226 }
|
|
5227
|
|
5228 /*
|
|
5229 * Return TRUE if a word appears in the list of banned words.
|
|
5230 */
|
|
5231 static int
|
|
5232 was_banned(su, word)
|
|
5233 suginfo_T *su;
|
|
5234 char_u *word;
|
|
5235 {
|
|
5236 return !HASHITEM_EMPTY(hash_find(&su->su_banned, word));
|
|
5237 }
|
|
5238
|
|
5239 /*
|
|
5240 * Free the banned words in "su".
|
|
5241 */
|
|
5242 static void
|
|
5243 free_banned(su)
|
|
5244 suginfo_T *su;
|
|
5245 {
|
|
5246 int todo;
|
|
5247 hashitem_T *hi;
|
|
5248
|
|
5249 todo = su->su_banned.ht_used;
|
|
5250 for (hi = su->su_banned.ht_array; todo > 0; ++hi)
|
|
5251 {
|
|
5252 if (!HASHITEM_EMPTY(hi))
|
|
5253 {
|
|
5254 vim_free(hi->hi_key);
|
|
5255 --todo;
|
|
5256 }
|
|
5257 }
|
|
5258 hash_clear(&su->su_banned);
|
|
5259 }
|
|
5260
|
|
5261 static int
|
|
5262 #ifdef __BORLANDC__
|
|
5263 _RTLENTRYF
|
|
5264 #endif
|
|
5265 sug_compare __ARGS((const void *s1, const void *s2));
|
|
5266
|
|
5267 /*
|
|
5268 * Function given to qsort() to sort the suggestions on st_score.
|
|
5269 */
|
|
5270 static int
|
|
5271 #ifdef __BORLANDC__
|
|
5272 _RTLENTRYF
|
|
5273 #endif
|
|
5274 sug_compare(s1, s2)
|
|
5275 const void *s1;
|
|
5276 const void *s2;
|
|
5277 {
|
|
5278 suggest_T *p1 = (suggest_T *)s1;
|
|
5279 suggest_T *p2 = (suggest_T *)s2;
|
|
5280
|
|
5281 return p1->st_score - p2->st_score;
|
|
5282 }
|
|
5283
|
|
5284 /*
|
|
5285 * Cleanup the suggestions:
|
|
5286 * - Sort on score.
|
|
5287 * - Remove words that won't be displayed.
|
|
5288 */
|
|
5289 static void
|
|
5290 cleanup_suggestions(su)
|
|
5291 suginfo_T *su;
|
|
5292 {
|
|
5293 suggest_T *stp = &SUG(su, 0);
|
|
5294 int i;
|
|
5295
|
|
5296 /* Sort the list. */
|
|
5297 qsort(su->su_ga.ga_data, (size_t)su->su_ga.ga_len,
|
|
5298 sizeof(suggest_T), sug_compare);
|
|
5299
|
|
5300 /* Truncate the list to the number of suggestions that will be displayed. */
|
|
5301 if (su->su_ga.ga_len > SUG_PROMPT_COUNT)
|
|
5302 {
|
|
5303 for (i = SUG_PROMPT_COUNT; i < su->su_ga.ga_len; ++i)
|
|
5304 vim_free(stp[i].st_word);
|
|
5305 su->su_ga.ga_len = SUG_PROMPT_COUNT;
|
|
5306 su->su_maxscore = stp[SUG_PROMPT_COUNT - 1].st_score;
|
|
5307 }
|
|
5308 }
|
|
5309
|
|
5310 /*
|
|
5311 * Turn "inword" into its sound-a-like equivalent in "res[MAXWLEN]".
|
|
5312 */
|
|
5313 static void
|
|
5314 spell_soundfold(slang, inword, res)
|
|
5315 slang_T *slang;
|
|
5316 char_u *inword;
|
|
5317 char_u *res;
|
|
5318 {
|
|
5319 fromto_T *ftp;
|
|
5320 char_u word[MAXWLEN];
|
|
5321 #ifdef FEAT_MBYTE
|
|
5322 int l;
|
|
5323 #endif
|
|
5324 char_u *s;
|
|
5325 char_u *t;
|
|
5326 int i, j, z;
|
|
5327 int n, k = 0;
|
|
5328 int z0;
|
|
5329 int k0;
|
|
5330 int n0;
|
|
5331 int c;
|
|
5332 int pri;
|
|
5333 int p0 = -333;
|
|
5334 int c0;
|
|
5335
|
|
5336 /* Remove accents, if wanted.
|
|
5337 * We actually remove all non-word characters. */
|
|
5338 if (slang->sl_rem_accents)
|
|
5339 {
|
|
5340 t = word;
|
|
5341 for (s = inword; *s != NUL; )
|
|
5342 {
|
|
5343 #ifdef FEAT_MBYTE
|
|
5344 if (has_mbyte)
|
|
5345 {
|
|
5346 l = mb_ptr2len_check(s);
|
|
5347 if (SPELL_ISWORDP(s))
|
|
5348 {
|
|
5349 mch_memmove(t, s, l);
|
|
5350 t += l;
|
|
5351 }
|
|
5352 s += l;
|
|
5353 }
|
|
5354 else
|
|
5355 #endif
|
|
5356 {
|
|
5357 if (SPELL_ISWORDP(s))
|
|
5358 *t++ = *s;
|
|
5359 ++s;
|
|
5360 }
|
|
5361 }
|
|
5362 *t = NUL;
|
|
5363 }
|
|
5364 else
|
|
5365 STRCPY(word, inword);
|
|
5366
|
|
5367 ftp = (fromto_T *)slang->sl_sal.ga_data;
|
|
5368
|
|
5369 /*
|
|
5370 * This comes from Aspell phonet.cpp. Converted from C++ to C.
|
|
5371 * TODO: support for multi-byte chars.
|
|
5372 */
|
|
5373 i = j = z = 0;
|
|
5374 while ((c = word[i]) != NUL)
|
|
5375 {
|
|
5376 n = slang->sl_sal_first[c];
|
|
5377 z0 = 0;
|
|
5378
|
|
5379 if (n >= 0)
|
|
5380 {
|
|
5381 /* check all rules for the same letter */
|
|
5382 while (ftp[n].ft_from[0] == c)
|
|
5383 {
|
|
5384 /* check whole string */
|
|
5385 k = 1; /* number of found letters */
|
|
5386 pri = 5; /* default priority */
|
|
5387 s = ftp[n].ft_from;
|
|
5388 s++; /* important for (see below) "*(s-1)" */
|
|
5389
|
|
5390 /* Skip over normal letters that match with the word. */
|
|
5391 while (*s != NUL && word[i + k] == *s
|
|
5392 && !vim_isdigit(*s) && strchr("(-<^$", *s) == NULL)
|
|
5393 {
|
|
5394 k++;
|
|
5395 s++;
|
|
5396 }
|
|
5397
|
|
5398 if (*s == '(')
|
|
5399 {
|
|
5400 /* check alternate letters in "(..)" */
|
|
5401 for (t = s + 1; *t != ')' && *t != NUL; ++t)
|
|
5402 if (*t == word[i + k])
|
|
5403 {
|
|
5404 /* match */
|
|
5405 ++k;
|
|
5406 for (s = t + 1; *s != NUL; ++s)
|
|
5407 if (*s == ')')
|
|
5408 {
|
|
5409 ++s;
|
|
5410 break;
|
|
5411 }
|
|
5412 break;
|
|
5413 }
|
|
5414 }
|
|
5415
|
|
5416 p0 = *s;
|
|
5417 k0 = k;
|
|
5418 while (*s == '-' && k > 1)
|
|
5419 {
|
|
5420 k--;
|
|
5421 s++;
|
|
5422 }
|
|
5423 if (*s == '<')
|
|
5424 s++;
|
|
5425 if (vim_isdigit(*s))
|
|
5426 {
|
|
5427 /* determine priority */
|
|
5428 pri = *s - '0';
|
|
5429 s++;
|
|
5430 }
|
|
5431 if (*s == '^' && *(s + 1) == '^')
|
|
5432 s++;
|
|
5433
|
|
5434 if (*s == NUL
|
|
5435 || (*s == '^'
|
|
5436 && (i == 0 || !SPELL_ISWORDP(word + i - 1))
|
|
5437 && (*(s + 1) != '$'
|
|
5438 || (!SPELL_ISWORDP(word + i + k0))))
|
|
5439 || (*s == '$' && i > 0
|
|
5440 && SPELL_ISWORDP(word + i - 1)
|
|
5441 && (!SPELL_ISWORDP(word + i + k0))))
|
|
5442 {
|
|
5443 /* search for followup rules, if: */
|
|
5444 /* followup and k > 1 and NO '-' in searchstring */
|
|
5445 c0 = word[i + k - 1];
|
|
5446 n0 = slang->sl_sal_first[c0];
|
|
5447
|
|
5448 if (slang->sl_followup && k > 1 && n0 >= 0
|
|
5449 && p0 != '-' && word[i + k] != NUL)
|
|
5450 {
|
|
5451 /* test follow-up rule for "word[i + k]" */
|
|
5452 while (ftp[n0].ft_from[0] == c0)
|
|
5453 {
|
|
5454
|
|
5455 /* check whole string */
|
|
5456 k0 = k;
|
|
5457 p0 = 5;
|
|
5458 s = ftp[n0].ft_from;
|
|
5459 s++;
|
|
5460 while (*s != NUL && word[i+k0] == *s
|
|
5461 && !vim_isdigit(*s)
|
|
5462 && strchr("(-<^$",*s) == NULL)
|
|
5463 {
|
|
5464 k0++;
|
|
5465 s++;
|
|
5466 }
|
|
5467 if (*s == '(')
|
|
5468 {
|
|
5469 /* check alternate letters in "(..)" */
|
|
5470 for (t = s + 1; *t != ')' && *t != NUL; ++t)
|
|
5471 if (*t == word[i + k0])
|
|
5472 {
|
|
5473 /* match */
|
|
5474 ++k0;
|
|
5475 for (s = t + 1; *s != NUL; ++s)
|
|
5476 if (*s == ')')
|
|
5477 {
|
|
5478 ++s;
|
|
5479 break;
|
|
5480 }
|
|
5481 break;
|
|
5482 }
|
|
5483 }
|
|
5484 while (*s == '-')
|
|
5485 {
|
|
5486 /* "k0" gets NOT reduced */
|
|
5487 /* because "if (k0 == k)" */
|
|
5488 s++;
|
|
5489 }
|
|
5490 if (*s == '<')
|
|
5491 s++;
|
|
5492 if (vim_isdigit(*s))
|
|
5493 {
|
|
5494 p0 = *s - '0';
|
|
5495 s++;
|
|
5496 }
|
|
5497
|
|
5498 if (*s == NUL
|
|
5499 /* *s == '^' cuts */
|
|
5500 || (*s == '$'
|
|
5501 && !SPELL_ISWORDP(word + i + k0)))
|
|
5502 {
|
|
5503 if (k0 == k)
|
|
5504 {
|
|
5505 /* this is just a piece of the string */
|
|
5506 ++n0;
|
|
5507 continue;
|
|
5508 }
|
|
5509
|
|
5510 if (p0 < pri)
|
|
5511 {
|
|
5512 /* priority too low */
|
|
5513 ++n0;
|
|
5514 continue;
|
|
5515 }
|
|
5516 /* rule fits; stop search */
|
|
5517 break;
|
|
5518 }
|
|
5519 ++n0;
|
|
5520 }
|
|
5521
|
|
5522 if (p0 >= pri && ftp[n0].ft_from[0] == c0)
|
|
5523 {
|
|
5524 ++n;
|
|
5525 continue;
|
|
5526 }
|
|
5527 }
|
|
5528
|
|
5529 /* replace string */
|
|
5530 s = ftp[n].ft_to;
|
|
5531 p0 = (ftp[n].ft_from[0] != NUL
|
|
5532 && vim_strchr(ftp[n].ft_from + 1,
|
|
5533 '<') != NULL) ? 1 : 0;
|
|
5534 if (p0 == 1 && z == 0)
|
|
5535 {
|
|
5536 /* rule with '<' is used */
|
|
5537 if (j > 0 && *s != NUL
|
|
5538 && (res[j - 1] == c || res[j - 1] == *s))
|
|
5539 j--;
|
|
5540 z0 = 1;
|
|
5541 z = 1;
|
|
5542 k0 = 0;
|
|
5543 while (*s != NUL && word[i+k0] != NUL)
|
|
5544 {
|
|
5545 word[i + k0] = *s;
|
|
5546 k0++;
|
|
5547 s++;
|
|
5548 }
|
|
5549 if (k > k0)
|
|
5550 mch_memmove(word + i + k0, word + i + k,
|
|
5551 STRLEN(word + i + k) + 1);
|
|
5552
|
|
5553 /* new "actual letter" */
|
|
5554 c = word[i];
|
|
5555 }
|
|
5556 else
|
|
5557 {
|
|
5558 /* no '<' rule used */
|
|
5559 i += k - 1;
|
|
5560 z = 0;
|
|
5561 while (*s != NUL && s[1] != NUL && j < MAXWLEN)
|
|
5562 {
|
|
5563 if (j == 0 || res[j - 1] != *s)
|
|
5564 {
|
|
5565 res[j] = *s;
|
|
5566 j++;
|
|
5567 }
|
|
5568 s++;
|
|
5569 }
|
|
5570 /* new "actual letter" */
|
|
5571 c = *s;
|
|
5572 if (ftp[n].ft_from[0] != NUL
|
|
5573 && strstr((char *)ftp[n].ft_from + 1,
|
|
5574 "^^") != NULL)
|
|
5575 {
|
|
5576 if (c != NUL)
|
|
5577 {
|
|
5578 res[j] = c;
|
|
5579 j++;
|
|
5580 }
|
|
5581 mch_memmove(word, word + i + 1,
|
|
5582 STRLEN(word + i + 1) + 1);
|
|
5583 i = 0;
|
|
5584 z0 = 1;
|
|
5585 }
|
|
5586 }
|
|
5587 break;
|
|
5588 }
|
|
5589 ++n;
|
|
5590 }
|
|
5591 }
|
|
5592
|
|
5593 if (z0 == 0)
|
|
5594 {
|
|
5595 if (k && !p0 && j < MAXWLEN && c != NUL
|
|
5596 && (!slang->sl_collapse || j == 0 || res[j - 1] != c))
|
|
5597 {
|
|
5598 /* condense only double letters */
|
|
5599 res[j] = c;
|
|
5600 j++;
|
|
5601 }
|
|
5602
|
|
5603 i++;
|
|
5604 z = 0;
|
|
5605 k = 0;
|
|
5606 }
|
|
5607 }
|
|
5608
|
|
5609 res[j] = NUL;
|
|
5610 }
|
|
5611
|
|
5612 /*
|
|
5613 * Compute the "edit distance" to turn "badword" into "goodword". The less
|
|
5614 * deletes/inserts/swaps are required the lower the score.
|
|
5615 * The algorithm comes from Aspell editdist.cpp, edit_distance().
|
|
5616 * TODO: make this work with multi-byte chars.
|
|
5617 */
|
|
5618 static int
|
|
5619 spell_edit_score(badword, goodword)
|
|
5620 char_u *badword;
|
|
5621 char_u *goodword;
|
|
5622 {
|
|
5623 int *cnt;
|
|
5624 int badlen, goodlen;
|
|
5625 int j, i;
|
|
5626 int t;
|
|
5627 int bc, gc;
|
|
5628
|
|
5629 /* We use "cnt" as an array: CNT(badword_idx, goodword_idx). */
|
|
5630 #define CNT(a, b) cnt[(a) + (b) * (badlen + 1)]
|
|
5631 badlen = STRLEN(badword) + 1;
|
|
5632 goodlen = STRLEN(goodword) + 1;
|
|
5633 cnt = (int *)lalloc((long_u)(sizeof(int) * (badlen + 1) * (goodlen + 1)),
|
|
5634 TRUE);
|
|
5635 if (cnt == 0)
|
|
5636 return 0;
|
|
5637
|
|
5638 CNT(0, 0) = 0;
|
|
5639 for (j = 1; j <= goodlen; ++j)
|
|
5640 CNT(0, j) = CNT(0, j - 1) + SCORE_DEL;
|
|
5641
|
|
5642 for (i = 1; i <= badlen; ++i)
|
|
5643 {
|
|
5644 CNT(i, 0) = CNT(i - 1, 0) + SCORE_INS;
|
|
5645 for (j = 1; j <= goodlen; ++j)
|
|
5646 {
|
|
5647 bc = badword[i - 1];
|
|
5648 gc = goodword[j - 1];
|
|
5649 if (bc == gc)
|
|
5650 CNT(i, j) = CNT(i - 1, j - 1);
|
|
5651 else
|
|
5652 {
|
|
5653 /* Use a better score when there is only a case difference. */
|
|
5654 if (spelltab.st_fold[bc] == spelltab.st_fold[gc])
|
|
5655 CNT(i, j) = SCORE_ICASE + CNT(i - 1, j - 1);
|
|
5656 else
|
|
5657 CNT(i, j) = SCORE_SUBST + CNT(i - 1, j - 1);
|
|
5658
|
|
5659 if (i > 1 && j > 1 && bc == goodword[j - 2]
|
|
5660 && badword[i - 2] == gc)
|
|
5661 {
|
|
5662 t = SCORE_SWAP + CNT(i - 2, j - 2);
|
|
5663 if (t < CNT(i, j))
|
|
5664 CNT(i, j) = t;
|
|
5665 }
|
|
5666 t = SCORE_DEL + CNT(i - 1, j);
|
|
5667 if (t < CNT(i, j))
|
|
5668 CNT(i, j) = t;
|
|
5669 t = SCORE_INS + CNT(i, j - 1);
|
|
5670 if (t < CNT(i, j))
|
|
5671 CNT(i, j) = t;
|
|
5672 }
|
|
5673 }
|
|
5674 }
|
|
5675 return CNT(badlen - 1, goodlen - 1);
|
|
5676 }
|
307
|
5677
|
236
|
5678 #endif /* FEAT_SYN_HL */
|